Guided weapons: the development of mathematical models and computer simulations in Australia

By Brian J. Polomka and Alex Biggs

(National Library of Australia archive: http://purl.nla.gov.au/nla/pandora/guided)

Part I

Introduction

With the end of the Cold War, it is time to publish hitherto unknown contributions by Australian scientists to its successful conclusion. The story of the Joint Project with Britain which resulted in the building of the Woomera range and its complex instrumentation, together with facilities at Salisbury (near Adelaide in South Australia) has been told in the book Fire across the desert. What has not been told is the story of System Assessment Division (SAD), the division responsible for the evaluation of the guided missile firings, including the end result: the effectiveness of the weapons. This outcome was only possible by the innovative use of computer technology; a true scientific experiment because it had never been done before, and there were many who initially believed it could not be done.

This story was put together by Brian Polomka, a journalist/historian in DSTO (WRE) who was employed as a researcher for the Joint Project History Team, with a major input from Alex Biggs, a research scientist; both were involved in the work in SAD. Brian pulled together the threads from many contributors into a consistent and interesting story which focuses on the contributions of individuals, be they technicians, engineers, or scientists.

For the internet, the story has been split into three parts to reduce access time.

Out of sight and flying fast

Serious problems faced the British and Australian scientists who wished to understand the behaviour of the new guided missiles during the late 1940s. Unlike previous aeronautical developments, you could not send a test pilot up to put them through their paces; they were expensive one-shot devices. The UK had begun a crash program to build supersonic wind tunnels but these could not be completed overnight and needed their own research program. Good work was being done at RAE by a small group flying rocket propelled free flight models to investigate supersonic and transonic aerodynamics. Their leader, T.F.C. (Tom) Lawrence, an Australian aerodynamicist, had pioneered the use of "bonkers", small charges which introduced known perturbations into the rocket's flight, enabling its recovery aerodynamics to be observed and flight characteristics deduced.

Previously wind tunnels had been large low-speed affairs and, as aeroplanes pushed nearer the speed of sound during the war, designers realised that they were approaching unknown territory where the old rules no longer applied. Instrumentation systems were crude and could not be relied on to give the right answers to questions which the designers had not yet learned to ask. Great strides had been made during the war with control technology and feedback circuitry but much had been done under a cloak of secrecy and many engineers had little knowledge of the new technologies.

When Bill Boswell told young George Barlow in 1949 that LRWE would not be designing missiles and hence would not need to use analogue computers he was only repeating current wisdom. The design of the new guided missiles, with their esoteric technology, was seen as the preserve of a select group of specialists. Alex Biggs, one of the early trainees, describes what was expected of the Australians. "Basically the contribution required of Australia in the early days once the range was operational, was to fire the missiles according to the UK plan, make all the required measurements of the test missiles, both pre-flight and in flight, then bundle up the records and forward them to the appropriate authorities in the UK for evaluation. Australians were not in a position to contribute to the firing plan because they were not familiar with the basic objectives of the experiments; nor were they able to contribute to the evaluation of the results, because they had not been involved in the design of the missiles. Students of colonial history will not find this situation unusual. "(1)

In many ways the development of the Woomera ranges created more problems than it solved. At first the planners had thought that the ability to fire guided missiles over an appropriately instrumented range would provide all the information that was needed both to develop the weapons and to assure the ultimate customers, the armed Services, that the product met their requirements. In some aspects their perceptions were correct -- in others they were to prove wildly inaccurate. Ultimately the situation was transformed to the extent that the same UK authorities regarded WRE as a centre of considerable expertise on the guided missiles fired at Woomera.

For Boswell's bright young men in Test Vehicles Group the concerns of the planners were, at first, of peripheral interest only. They would often work to 3 am preparing an RTVl for firing, snatch a couple of hours sleep and then get back to the launcher to try to fire the missile before the heat of the day made optical tracking too inaccurate. When the missile failed to perform, as they frequently did, they became frustrated when all they could do was bundle up the records and send them off without making any input of their own as to the cause of failure. The main advantage of the Australian range, the recovery feature, meant that they could record that a wing had, say, torn off in flight. But why? Was the wing not strong enough or had the missile control system demanded more of the missile than it was designed to do?

These young men had passed through varied and thorough training in Britain. To give some idea of the different disciplines encompassed; at the time when George Barlow was working on simulators in Building 134 RAE; Lister, Deegan and Gillard were working at RAE (Bramshot) on telemetry; White, Keats and Higgs were involved with studies and assessment; Barnsley, also at Bramshot was working on RTVl structural and project matters and Geoff Lee (and Les Pearson?, Bob Weldon?) was at Westcott working on propulsion including RTVl propulsion. Others who were not so directly involved with RTVl included Ron Whitten at the Admiralty Signals and Radar Establishment, Peter Twiss at the National Physical Laboratory, Merv Kirkpatrick at the Radar Research and Development Establishment, Alan Sharpe at the National Gas Turbine Establishment, Bob Rockliff at the RAE Structures Department and Keith Thomson at the RAE Aerodynamics Department. When all these returned to Australia it would indeed have been surprising if they had been content merely to test and fire rockets and to send the records off to Britain without demur.

Learning about the new technology

While the trainees were learning the detail of the new technologies in the UK the secret wartime knowledge of the theory was being spread in Australia. A sympathetic Chifley Government had backed the Joint Project whole heartedly and had spent money and effort willingly in order to expand the scientifically based defence manufacturing capability that had been so inadequate when Labour had taken over the reins of government in 1940. This governmental support, or rather enthusiasm, for the Joint Project had enabled entrepreneurial scientists like W.A.S. Butement and Professor E.O. Willoughby of Adelaide University to institute training programs which bore fruit in years to come.

In Adelaide, a rustic city of little more than 300,000 people at the end of the war, Professor Willoughby seized his opportunity and produced, first, courses for graduate engineers in the new science of control technology and servomechanisms and then courses within the undergraduate curriculum which would ensure that students gained the skills needed to exploit the new technologies. To help the University do this the Department of Supply made regular grants to the Adelaide University Electrical Engineering Department to enable it to build analogue computing equipment. (2)

The initial idea for the UK traineeships had been to build up the necessary skills in Australia but there had also been an element of inducement thought necessary to entice young men to work in the remoteness of the desert. The first few received limited training in specific range related jobs but after that there was a deliberate policy of placing scientifically well qualified men in research areas where they would gain in-depth knowledge likely to be useful in a variety of not yet defined tasks. Trainees learnt entire new disciplines but they were also given an entree into British scientific establishments that was to prove of mutual benefit. The British intention to use Australia only as a testing ground did not survive long against a background of labour and manufacturing shortage in the UK and the obvious waste of highly skilled manpower in Australia.

The British had found the Australian 'trainees' to be of a much higher calibre than the word implied, as it was normally used in the UK for undergraduate level workers. They had been chosen from the best of young graduates offering and rapidly became valued members of the establishments to which they had been sent and from which they were often released only reluctantly. In early 1950 Boswell and Chuck Bayly reported back to Australia a plan to devolve more of the experimental work of the RTVl program onto Test Vehicles Group and to build at least some of the RTVls in Australia because the British contractors could not build enough.

Simulators an answer?

Messrs Bolshaw and Riddle, of the UK Ministry of Supply, came to Salisbury in October 1951 to discuss the details of the transfer of some of the RAE research program. Major Riddle recommended the acquisition of an RTVl simulator as a necessary adjunct to the envisaged research work. The UK formally agreed to provide a machine in the following January but there seems to have been a rather casual air, with nothing too well defined, about the arrangement. Treharne, Principal Officer of TV Group, proposed in April 1952, that the Control and Guidance Group should begin construction of a simulator. Building it would be only one of the necessary steps as the scientists who would use it would also have to be trained in the required disciplines.

Nothing was done at the time because of threats to TV Group's very existence with suggestions that aerodynamic research be done by another Department of Supply laboratory. In November 1952, J.P. (Jack) Lonergan and E.G. (Ted) Hayman, attached to the simulators group at RAE, issued two memos on the production schedule and power supplies needed for the analogue computing elements being built for Australia by Elliott Bros. of the UK. They had previously written a Tech Memo, GW.189, a "Preliminary Report on Australian Simulator". Following TV Group's successful bid to expand into the research area there was an exchange of visits between Wass and Gait of RAE and Collingwood and Treharne of LRWE. Charles Wass, head of the Dynamic Analysis Division and Jack Gait, an expert on simulation, accompanied by H.S. Stewart-Jones from the Ministry, came with a dual purpose, to give lectures on theoretical studies and simulators and to discuss some sharing of the RAE's research programme with the increasingly expert Australian missile teams.

At that time beam-riding guidance systems were being extensively investigated. In beam-riding a radar on the ground tracks the target and the missile tries to fly along the centre of the beam to intercept the target. Homing guidance was favoured for the next generation of missiles but it was by no means clear which would be the best of several possible navigation laws for homing systems to follow. In homing guidance the missile is much more intelligent, carrying its own sensor to track the target and is able to fly towards the future position of the target. The UK visitors agreed that a part of their research programme into new guidance laws could be done at Salisbury.

There was another advantage for RAE in the new arrangements. Harry Bateman, the energetic and shrewd DDGW(A) was less than confident that the private firms doing work under contract for the UK government were completely frank and open about all they discovered in the course of their work. LRWE was another government establishment. The young men who would be doing the work were well known professional colleagues who had no commercial or other reasons to keep secret what they discovered. It meant that work could be shared confidently with the spur of friendly rivalry to act as an incentive to achieve excellence.

Simulators would obviously be essential if this work was to be done in Australia but the size and form still had to be decided. Initially the British intended to provide components only, amplifier modules and similar assemblies. These could be added to existing contracts and would require the minimum of design work. In the meantime it was agreed that the Australians would push ahead with a locally constructed simple single plane RTVl simulator. Charles Wass, responsible for the simulator and studies group at RAE gave J.P. (Jack) Lonergan the job of specifying just what computing equipment the British government would supply to Australia. Lonergan enlisted the help of E.G. (Ted) Hayman and Alex Biggs so that a representative Australian view could be put forward. At the time, Biggs was fully involved in building a radar simulator to rehearse the UK's first interception by a beam rider of an aircraft in flight at the Aberporth (Wales) range.

Lonergan, who was also liaising with Elliott Bros, felt that what the Australians on the spot asked for they would probably get. He had argued for the provision of a complete machine rather than just some of the computing elements. At one crucial meeting, when it became clear just how much work would be involved, Lonergan unexpectedly found himself in charge of the project. Elliotts at Boreham Wood were in desperate straits with TRIDAC and said they could do no more than extend their production runs for standard items. Ken Simmonds, Chief Engineer at Elliotts Lewisham instrument factory offered to build the Australian simulator if Lonergan took over as design authority, responsible for the whole system concept. In the event, Simmonds' proposal was accepted and Lonergan found himself working at Lewisham and later at the Chatham works.(3) Meanwhile Wass' own simulator (DYADIS) at RAE was held up owing to the illness of the SPSO in charge and he requested that Alex Biggs stay in UK be extended to take over and complete the task.

Simulators at that time were basically analogue computers but often included large chunks of real hardware (such as hydraulic fin actuators). One of the first to recognize the need for Australia to become involved with simulation had been C.P. (Phil) Gilbert, an English electronics engineer, who had been recruited to work at the LRWE in the Test Vehicles Group headed by Chuck Bayly, also an Englishman from ICI. Gilbert was an expert in the new field of servomechanisms and control theory and was well placed to apply it to analogue computer development. At this time the transistor was still largely a laboratory curiosity. Everything depended on thermionic valve (vacuum tube) technology. Gilbert was given the job of designing a simulator, named ARTVS (pronounced ARTUS), for Australian RTV Simulator.

He visited the UK in early 1953 and afterwards completed the design of a computing amplifier which used some of the war surplus valves that LRWE had in store in the many thousands. Next came the task of trying to get it and the necessary associated equipment built. The first two production amplifiers were finished by the end of March and workshops promised a further 12 a month. Gilbert found it hard to get work done at a time when LRWE was expanding so rapidly. Electronics Group was too busy to design and build a photo-electric curve follower and suggested Engineering Development Group. They, in turn, said there was too much experimental work needed. Groups found it easier, in many cases, to build their own equipment. The result of this shortage of both men and material during the 1950s was eventually to lead to the presence of an engineering group within SAD which had a significant influence on the quality of equipment produced.

Operational amplifiers - how were they different

In what way was the computing amplifier different from other amplifiers? Before World War 2 the main use for amplifiers was in radio and sound amplification equipment. These were not demanding applications. Amplification did not have to be exact, treble or base controls compensated for lack of linearity. Volume controls adjusted the sound to suit the listener. During the war engineers found they could replace some of the mechanical elements of anti-aircraft gun directors/target predictors with operational amplifiers.

These were so called because they performed various mathematical operations. They found that the application of negative feedback around a high gain DC amplifier would produce a circuit with a precise gain characteristic that depended only on the feedback used. The input current may be neglected and the input voltage may be considered zero, i.e. the input point is a virtual earth'. By the proper selection of feedback components, operational amplifier circuits could be used to add, subtract, average, integrate and differentiate voltages. The very high gain of the amplifier meant that, ignoring internal drift, any errors which it might contribute to the resistor network would be minute and could be ignored.

Effectively the voltage analogues of physical quantities were manipulated by the amplifier and its resistor/capacitor networks. Therein lay both the strength and the weakness of the analogue computer. It was as fast as the circuit response. Real-time computation was easy for most missile work and faster or slower than real time computation possible by selecting different values of the computing elements. Unlike the serial digital processors of the day the analogue machines worked in parallel as outputs from one section of the model instantly became inputs to other sections.

The weakness lay in the fact that the accuracy of any computation depended on the precision of the controlling components, the resistors and capacitors and the precision of the voltage supplies to the amplifier itself. For much of the early missile simulation work this was not a problem. The computing elements were, in general, accurate to about 0.3%. Many of the variables such as air pressures, temperatures, missile velocities, positions, etc. could not be measured to anywhere near this degree of precision. The performance parameters too, of the missile, could not be specified precisely.

The technique of mathematical modelling made possible by the new analogue computers proved a very powerful means of investigating not very well known phenomena particularly where many variables were involved. The concept of modelling was not new. Basically a mathematical model is a mathematical description of something so organized that it can be manipulated to explore facets of the modelled object's behaviour. Mathematics had long been used, for example, to describe planetary orbits. The development of computers enabled extremely complex systems to be modelled for the first time and their complicated behaviour patterns to be explored and explained.

The question then arose; how believable was this information from the model? The essence of the development at WRE of mathematical models was the digestion of huge amounts of trials data (millions of data points), to verify and correct the models. Having verified that the model was an accurate enough description of the real system, over a range of conditions, one could then use the model, with confidence, to predict the behaviour of the real system under conditions when it would be impossible or too expensive to test the real system. The technique is a general one, not confined to guided weapons but to applications as diverse as biological (living) systems and economic systems.

ARTVS - Australian RTV Simulator

An early photo of ARTVS shows three, six foot Post Office equipment cabinets partly filled with power supplies, regulators, a two-pen recorder and some 24 plug-in amplifiers all bolted onto a wheeled raft. Many of the panels had circular holes still awaiting the insertion of meters and the plug-in amplifiers were fitted with little perspex fronted boxes with 'rat-trap' fuses inside. A jumble of cables such as telephonists then used on manual exchanges connected the amplifiers together. Each amplifier was fitted with tag boards which enabled it to be converted to an integrator, adder or inverter as required.

Control was simple as relays earthed all inputs and integrating capacitors in the quiescent state. When the 'Initiate' switch on the main control panel was closed all the relays were energised and the computation began. It hardly looks impressive enough to be the tool of a research revolution but by March 1954 Gilbert was able to report that single plane simulation of a beam rider had been achieved and the equipment was being expanded to enable three dimensional simulation to be undertaken. Other equipment such as limiters, squarers, electro mechanical sin/cosine units, etc. had either been made or was being developed.

TRIDAC and AGWAC

Meanwhile, in the UK, what was probably the world's most ambitious analogue computer project was being assembled at the Royal Aircraft Establishment by Elliott Bros. It was called TRIDAC (Tri-dimensional analogue computer) and it required a two storey building to house it. RAE engineers had designed the excellent high gain and very stable amplifier which had then been engineered by Elliotts. These they had built into solid steel cabinets (the smaller AGWAC was referred to in press reports as the 32 ton AGWAC) but what also took up a great deal of the space in the big building was the the high pressure oil system. This was designed to drive the servomechanisms so that they could respond very rapidly to allow TRIDAC to simulate missile responses to frequencies as high as 100 Hz.

With that flexibility of management for which the British used to be famous the TRIDAC contract was extended to provide a smaller but still advanced analogue computer for Australia. It would be known as AGWAC (for Australian Guided Weapons Analogue Computer) and would not have the the oil driven servos of TRIDAC but, as the Australians later found out, their electrically powered servos which could respond to perhaps 10 Hz were adequate for most simulations.

The Kent Messenger of 10 October 1954, carried a story about both machines headed, "Kent-made robot brain for rocket experiments" and with a sub-heading saying, "SAVES THOUSANDS OF POUNDS". By this time AGWAC had left the UK and was being installed and tested in Australia by Elliott engineers assisted by Jack Lonergan and Tom Buckland, a technician who had earlier joined Lonergan at Elliott's factory to learn how to maintain it. Components of TRIDAC were on show at Farnborough.

The story made two points worth noting. The first was in the sub-heading. Both machines would indeed save many thousands of pounds and their introduction marked a turning point in the testing of expensive rocketry and, ultimately, in the testing of many other things as well. The second concerned the production of the electronic components in Elliotts Rochester factory. Previously this had only been used to produce mechanical components and quite "raw" labour- -men and women- -was recruited from the Medway area to help build the new electronic equipment. The Elliott built mechanical servomechanism computing elements which were used in both AGWAC and TRIDAC were, at their best, superb examples of the metal machinist's art. The days of such men were numbered with the introduction of the new electronic 'brains' which needed skilled designers and operators but which were infinitely more flexible in operation and could be built largely with unskilled labour.

Already in the early fifties the missiles being tested at Woomera were using the most advanced technology then available and understanding their behaviour in-flight and recognising abnormal or normal behaviour, had gone beyond engineers poring over telemetry records and doing slide-rule calculations. Indeed the development of those missiles in the UK had gone hand-in-hand with the development of computers to simulate their flight behaviour in the laboratory. Basically a missile in flight received electronic guidance signals, which were processed internally to generate movements of its control surfaces which, in turn, generated the aerodynamic forces necessary to steer the missile to its target.

If by some means, one could generate the guidance signals in the laboratory, process them the same way and apply them to a real control surface actuator set up in a rig in the laboratory, one could generate a time-history of fin movement broadly similar to that achieved in flight. Early simulators did just that.

The new weapons and a new role for the simulators?

Simulation was obviously useful in helping understanding of the new weapons and their dynamic behaviour but by the mid l950's some of the engineers and scientists concerned were seeking an expanded role for their analogue computers. The first Guided Weapon Acceptance Trials to take place were those of the "Blue Sky" - renamed "Fireflash" as a Service weapon.(4) The December 1954 WRE Progress Report stated that these were to begin in July 1955 and that 82 weapons would be fired. This represented a large commitment, as the trials were elaborate, and demanded careful positioning and control of both attacking and target aircraft to produce the conditions demanded by the basically statistical nature of the trials design. The statistical design simply sampled points within the 'operational envelope' of the weapon - an envelope with upper and lower limits on height, speeds of attacking and target aircraft, angle between aircraft tracks, acceleration of aircraft and possibly other parameters. Inevitably, the question arose - "Why a statistical approach? Why 82?." Further, 'Fireflash' was the first of a whole sequence of weapons. Clearly, acceptance trials were likely to demand much more effort than any other trial series.

People at WRE in Trials Division under Stan Price and within Flight Research Division under Tom Lawrence were beginning to ask questions about the way acceptance trials should be conducted. This began in Trials Division and in April 1955 Assessment Group was set up there to assess "trials, weapons and range performance". Bill Watson, who was very familiar with the vast resources a statistical approach demanded in the much simpler area of Bomb Ballistics and was searching for alternatives was chosen to lead it. Bruce Moon, a young New Zealander and a trainee during 1952/54, played a major role in those early debates. During his traineeship he had been very impressed with the work of W.H.B. Cooper at the Royal Radar Establishment, Malvern. Cooper's approach to the development of weapons differed in emphasis from that of the Royal Aircraft Establishment, Farnborough. Cooper was concerned to compile mathematical equations, representing the dynamic approach of a missile to a target, with careful pre-flight measurements of those imperfections which caused a missile miss a target. Moon argued that this method could be adapted to Acceptance Trials.

Conflict between establishments

A debate with UK followed. The debate was sharpened by RAE/RRE competitiveness and by the increasing confidence, with nationalistic overtones, of Boswell's 'young men'. On 21 April 1955, A.V.M. McGregor, the Director of Guided Weapons Trials in UK, acknowledged the validity of WRE involvement in trials design by inviting WRE to provide an Assistant to the Officer in Scientific Charge of the next series of evaluation trials, those of Blue Jay, called 'Firestreak' when it went into use as a Service weapon. Controller Brown responded by saying he would send Watson and Moon to the UK to discuss Blue Jay evaluation and Moon could stay there for a time.(5) This did happen in July 1955. In fact, Moon stayed there for over twelve months in a controversial atmosphere, as he was criticised by the UK for spending too much time at RRE with Cooper and Briggs instead of at RAE with Gerard, the man responsible for acceptance trials design and analysis in RAE. Gerard was firmly committed to a statistical approach. The question of Moon's status was also touchy. Was he a co-sponsor or an 'Assistant Scientific Officer'?

The model approach found a very sympathetic hearing within Flight Research Division under Tom Lawrence,(6) where modelling already underpinned the whole investigation of RTVl aerodynamics, control and beam-riding. The knowledge and equipment of that Division would be needed if there was to be any attempt to implement such an approach to acceptance trials. In August 1955 Trials Assessment Group was transferred to the newly-named Systems Assessment Division - and thereafter Tom Lawrence became the major advocate of the approach.

The debate was mainly with Gerard of RAE. A compromise was proposed by Cooper and Briggs of RRE that 10% of the Blue Jay weapons should be allocated to special experiments for modelling. This proposal was expounded in WRE by Watson and was well received in WRE by Boswell, Lawrence and Price (6-9-1955). Ultimately the scheme accepted by DGW (now Air Commodore B.A. Chacksfield) on 23 March 1956 at a meeting in Mr Boswell's office, with W.B. Gerard in attendance, was that "a model experiment involving about 10 rounds should be staged in parallel with the planned statistical experiment and the results of both tests examined before the second phase of the trials began." This was a wise decision, as it would not be easy to build a model of a weapon, whose Research and Development had not allowed for such.

The debate on other weapons was less vehement, partly because John Mercer replaced Gerard at RAE and Lawrence and Mercer found large areas of agreement. Visits to WRE by Cooper and Mercer followed. During 1956 Alex Biggs visited UK to discuss the planning of Red Duster Acceptance trials with Mercer and Cawthorne at RAE. Dick Cawthorne was the UK nominee for the position of Officer in Charge (Scientific). Les Witchard also visited the UK to report on preliminary Blue Jay studies. When Bill Boswell visited USA on the first inspection of US Ranges in September 1956 he included Bill Watson specifically to try to discover if moves were being made along the same lines and, in fact, the beginnings of the same approach were found at White Sands proving ground.

Alex Biggs still remembers those discussions as they planned the firings and mathematical modelling needed for the evaluation of Bloodhound.(7) "In essence what we proposed to the UK authorities was that the trials be planned primarily to validate a mathematical model of the missile and that the bulk of the answers sought be subsequently obtained from the model rather than from the real life trials. WRE would prepare the model with help from the contractors, and validate it against the trials results. This proposal was seen in UK to be something of a gamble because previous attempts to achieve a one-to-one correspondence between computer simulations and trials results had not been conspicuously successful. The Americans, it appeared, were not doing it that way, but they could afford to fire many more missiles.

"The Australian proposal was almost killed at birth by the UK insistence on a package deal whereby we would do both Bloodhound and Thunderbird, whereas we knew we had barely enough manpower and computer resources to do the one we proposed, Bloodhound. Fortunately we were let off the hook when the UK decided not to proceed with Thunderbird I so we were able to concentrate our resources on Bloodhound."

The model approach becomes policy

The culmination of all this exploration and advocacy was a letter which Air Commodore B.A. Chacksfield sent to UK Guided Weapon contractors dated 8 January 1957 stating:

"Anglo-Australian Guided Weapon Acceptance Trials

"You will be aware that current thinking within the MOS on the design of the Red Shoes (8) Acceptance Trials envisages a basic and fundamental use of a mathematical model of the missile system. The method of attack may be sub-divided into five phases -

1. The generation of the quantitative information required to set up a useful simulator model.

2. The setting up of the model.

3. The checking of this model by comparison between the results of Acceptance Trials Firings and those forecast by the model.

.4. The correction of the model if in the checking phase it proves to be inadequate.

5. The use of the resultant simulator model to generate the answers to the operational questions posed."

"The present situation is, therefore, that -

(a)RAE and WRE will be jointly responsible to me (DGWT) for recommending the design of Acceptance Trials.

(b)WRE will take responsibility for building up and checking against Acceptance Trials Firings the simulator model referred to above.

In order to discharge this responsibility it will be necessary for your firm to provide, by the time Acceptance Trials begin, sufficient information to allow an adequate simulation to be attempted. The Acceptance Trials Section in GW Dept., RAE will be responsible for advising me that the necessary steps are being taken to provide this information, for coordinating it and for passing it to the appropriate part of WRE.

(c) If in the checking phase the available model should prove inadequate then WRE will be responsible for undertaking such work as is necessary to correct it. In this work WRE would need your sympathetic assistance for carrying out such laboratory work as is necessary. This might possibly include special live firings of the weapon.

In these circumstances demands to be placed on your firm are -

1. To produce initial information for the setting up of the simulator model.

2. To give sympathetic assistance should extra work prove necessary to correct the model.

"On the first of these, members of the RAE GW Dept. Acceptance Trials Team are already in close contact with your firm and are gathering with the full cooperation of your staff, the information needed. This process will doubtless continue and be augmented by visits from key officers of WRE to the UK. The need under item 2. cannot be defined with any precision at this time. It is perhaps sufficient to say that a major discrepancy between the simulator and the result of the Acceptance Trials Firings would be viewed as seriously as a persistent mechanical fault, and the corrections of the model would accordingly call for urgent work by all the parties concerned.

(Signed)B.A. CHACKSFIELD Air Commodore

Director of Guided Weapons (Trials)" (9)

Part II

References
1. Tom Lawrence challenges this perception. It may be a matter of personal perspective. Certainly the initial planning called for trainees to be selected to do just the sort of work Alex describes but by 1949 Pye is on record as wanting the very best candidates to be selected and given, not specific rocket range training, but specialist experience which would suit them for as yet unspecified scientific work.
2. Based on discussions with the late L. J. Dunne as to reasons for the excellence of WRE work so soon after establishment.
3. Meeting at RAE, 12/12/52. TM22
4. Fireflash was the first British guided missile to be fitted to fighters. Hitherto they had attacked aerial targets with machine guns or cannon.
5. Folio 1 File SA5407/2/1 Trials Policy - Acceptance Trials - General
6. See Lawrence letter of 5 May, 1986 where he mentions, among other things, his wartime experience in evaluation trials of aircraft where the mathematics/physics of the system, rather than a statistical approach, had been paramount.
7. Initially called Red Duster
8. The code name for Thunderbird 1
9. The above short history of the beginnings of simulation as applied to Acceptance Trials was supplied by Bill Watson.

AGB COMPUTER ART home page