The organisation of the Domestic computer office described here developed gradually as the computer progressively played an increasing part in the work of the aircraft design office. The present form of the organisation has been established for the last two years, previous to this there was a period of about 18 months during which the computer office was gradually building up. The organisation has in effect been tailored to effectively serve the Firm in which it exists and the test of time has shown it to be efficient and to maintain a high level of productivity. In this latter respect much credit is due to the makers of the computer for their good engineering and to the Servicing Engineers, Programmers and Tape Editing Staff of the Computer Office for their enthusiasm and professional competence.
As to the organisation itself, there are other forms which could have been laid down, but at the time of acquiring the computer, there was no precedent to be guided by and in any case the initial utilisation of the computer was insufficient to require the sub-dividing of the computer office into separate functions. There was also the question of programme development, very few programmes existed in the initial stages and therefore, most of the effort was of necessity concentrated on writing programmes and trying them out on the computer The amount of production i. e. actual computing, was small in comparison to what it is today and the Tape Editing Staff as such did not exist.
Increased delegation of authority was introduced in stages as the variety and volume of work increased until eventually a settled organisation evolved, but it is not pre-supposed that any future domestic computer organisation engaged on similar work would resolve into the same pattern. A lot depends on the computer itself; the very latest computers are more amenable to auto-coding and it is possible that a programming team as such may disappear; Engineers and Physicists concerned with creating the problems may also become responsible for their own programming.
The Computer Office is functionally required to provide a service to the aircraft design team so that the day to day calculations and mathematical work are taken out of the hands of the Engineers and Physicists who can then spend more time on creative thinking. This service can be quite extensive in scope if the Computer is used as a design tool rather than simply as a high speed desk calculator turning out rapid solutions to set problems. As a design tool, the computer can greatly extend the scope of a technical investigation by repetitive use on a wide variety of data. The design process of an aeroplane is one of successive elimination and compromise to evolve the optimum design to meet specified performance requirements. Design conclusions are reached by assessing the evidence of calculations relating to performance, shape, strength, stiffness, weight and controllability. Calculations may be repeated many times in the light of previous results and therefore, rapid availability and distribution of information throughout the design team is desirable. For example, take the case of a wing stressing programme. The use of the programme need not be confined to a stress analysis of a finalised structure, but may be used to help in the design of the structure by repetitive use on systematic variations of such parameters as: number of spars, the disposition and angle of sweepback of spars, cross-sectional properties of spars and skin thickness. From the evidence of these repeated calculations it is possible to evolve the lightest structure to satisfy the loadings. In addition much useful information on wing deflections and twists emerges as a by-product for use by the aeroelastic section of the design office.
The design team consists of several groups, each specialising in a certain aspect of design, such as aerodynamics, performance predictions, stressing, aeroelasticity, weight estimation, control and stability and pressurisation. None of these groups can exist alone, they are all in varying degrees influenced in their decisions by the dictates of design requirements of the other groups. Problems requiring computation are passed into the Computer Office from all groups, the computed information in typewritten form is distributed not only to the particular group concerned with the problem, but also to other groups whose work is directly influenced by these results. In this respect the Computer Office acts as a clearing house for information and detailed records are kept of all transactions.
The diagram (Fig. 1) illustrates the organisation established within the computer office in order to serve and assist the aircraft design office.
The Computer Office comprising Programmers, Tape Editing Staff and Maintenance Engineers is regarded as a function of the Electronics Laboratory and is therefore, presided over by the Head of Electronics. The Chief Programmer controls the Programmers and Tape Editing Staff and is assisted by Section Leaders each specialising in certain aspects of the work.
The Maintenance Engineers are not responsible to the Chief Programmer, but come under the direct control of the Head of Electronics. With this arrangement the Engineers can do their job without being subjected to too much pressure from the Programmers.
The Programming Staff is divided into three sections so that some degree of specialisation can exist. The three sections are identified by the type of problems they each deal with, namely:- Aerodynamics, Stressing and Miscellaneous problems, (which includes Data Processing).
Each section consists of 3 Programmers and is controlled by a Section Leader who is responsible to the Chief Programmer. Programmers may be transferred from one section to another according to work priorities. The Tape Editing Section consists of four Computer Clerks, directly supervised by a Section Leader who is again responsible to the Chief Programmer.
The supervisory duties of the Chief Programmer and the Section Leaders include the following:-
There are three different classes of work in the Computer Office; Programmings Tape Editing and Computer Maintenance. The qualifications, experience and training of the Personnel involved are now discussed individually:-
The Programming Staff consists of a nucleus of Mathematicians who are responsible for designing Programmes for all problems requiring solution by the Computer. The originator of a problem is first required to write a detailed mathematical description of the problem so that the Programmer can get a complete understanding of what the originator is aiming to do. The Programmer then proceeds to reduce the problem into computer steps, organise the performance of the computer according to its facilities and storage, then prepare the coding sheet.
It has been found that Honours Degree Mathematicians are desirable for this work and that they are at their best when recruited straight from University while they are still very familiar with Mathematical tricks. It usually takes a mathematician about four months time to be able to make a reasonable job of programming the computer, although it is possible after only the first four to six weeks to carry on without direct assistance, but with the aid of a Programmer's Handbook. The computer is a very reliable checker and will quickly sort out any mistakes.
The Programmers individually use the computer for checking and developing the programmes which when completed are stored in the Programme Library ready for use when computations are required.
Girls with G.C.E. standard of education, in Mathematics and English, have been found to be capable of this work, in fact for tape editing alone, they are above the required level. Tape Editing, however, can lead to computer operation after a little experience and G.C.E, standard is desirable for this work. In addition to tape-editing, the girls are responsible for keeping records of all Computer work. The filing system involved in the Tape Editing Room can become quite complex, for instance, any one programme may be used thousands of times, each time with a different number tape. The number tapes must be identified in relation to the state of development of an aeroplane at the time so that future reference can be made in the event of uncertainty. To complicate this, the programme itself may undergo some revision in the light of new theories in the formulation of the problem and it may also be used on different aeroplanes. A weekly .statement is issued on work done and needless to say9 the Computer itself is employed for this work.
The most important aspect of Computer operation is ensuring that the correct information is fed into the computer. Unfortunately there appears to be an adverse psychological aspect relating to Computer Data; the people concerned with the preparation of data and also the people concerned with tape editing are apt to be less careful than they would have been if they were solving the problem by long-hand methods. Careful control has to be exercised to ensure that checking procedures are adhered to.
Computer Data is received from the Technical Offices i.e. Aerodynamics, Stress etc., in the form of tabulated sheets of numbers relating to physical properties of the aeroplane. The Computer Staff have no means of verifying that these numbers are all correct and a system of checking has been established in each respective Technical Office.
The Tape Editing Staff prepare the tape on Keyboard Perforators from the numbers listed on the tabulated sheets. In general two tapes are prepared by different operators and then compared either automatically or manually to check for punching errors. Alternatively, a reprint can be obtained by feeding the prepared tape into teleprinter equipment, this reprint is then visually compared with the original tabulation.
If these procedures are faithfully adhered to, one can feel confident that no tape editing errors will go undetected.
An overall check can be employed when practicable by check summing in the Computer, this also verifies that the input tape reader on the computer is functioning properly.
It has been found desirable to have at least one technician in the maintenance team plus two or three practical engineers with experience equivalent to Radar Mechanics. For a man with no previous experience of digital computers it takes approximately six months to become sufficiently familiar with the machine. Unfortunately, he cannot be learning all the working day because when the computer is in action it is unwise to allow him to interfere with the circuits. It is only when faults appear that he gets any real practice. Of course, much can be learned during the preventative maintenance period each day about the normal process of adjusting wave forms etc., but usually the Programmer's are needing the machine long before the Engineers are prepared to hand it over.
The idea of Preventative Maintenance is to try to anticipate trouble and then to take the necessary safeguarding action, in the broadest sense this involves Marginal Testing and the systematic checking of circuit elements throughout the Computer.
Marginal Tests are carried out by adjusting the conditioning potentials of valve circuits by a small percentage from normal and in this condition, test programmes are fed into the machine and it's performance observed. If any malfunctioning occurs, the cause has to be tracked down and any suspect components replaced. It takes approximately 1/2 hours to complete these tests even when everything is satisfactory. Circuit checking is carried out after Marginal Tests are completed, a small section of the computer is dealt with for about 1 hour each day to check if there is any deterioration in components and circuitry,, It takes several weeks to completely work round the whole computer.
Three hours per day are set aside for Marginal Tests and Circuit checking. The Engineers commence work 1 hour earlier than the Programmers, therefore, only two hours of the Programmers day is affected. The computer time is divided in the following proportions:-
Although there may appear to be a disproportionate amount of time allocated to the Engineers, it has been found profitable to adopt this daily maintenance routine. A very high standard of operational reliability has resulted and the consequent reduction in abortive machine time during computations easily off-sets the time spent by the Engineers.
The Chief Programmer prepares a daily time table for the computer based on degree of priority of the different jobs. During the course of the days there is a continuous flow of work onto the Computer and the Tape Editing Staff are kept busy not only preparing tapes but also operating the teleprinter equipment used in producing typewritten results from computer output tapes.
Machine time is apportioned between Development time and Production time. Development time is the time used by Programmers when trying out their Programmes on the Computer. Production time is the time absorbed when actually doing computing. Production may be carried out by Programmers or by a member of the Tape Editing Staff according to the degree of complexity of the Programme.
The working day for the Programmers and Tape Editing Staff is divided into two shifts:-
1) 8.24 a.m. to 5.00 p.m.
2) 1.15 p.m. to 9.36 p.m.
Most of the staff are on the first shift and a limited number are on second shift, the allocation of time for development and production is to some extent settled by staff availability.
In general 10.30 to 12.30 is used for development because the majority of the staff are there to carry out their own development work. Production is done in the evening (5.00-9.36) when only one or two people are on duty. The afternoon is divided into development and production according to the degree of priority on each job.
In practice the Computer timetable is continually adjusted throughout the day for the following reasons:-
A phrase often used by Computer Salesmen, is that the appetite grows with eating. From my personal experience of computer application, I can readily agree with this reasoning, for although we did anticipate a justifiable volume of work when our computer was installed, it has been most surprising and revealing to discover the extent to which the computer has exercised its influence. As machine consciousness spread throughout the Aircraft Design Office after a few preliminary but powerful demonstrations of its possibilities had been made, many unforeseen applications arose. The variety of problems dealt with by the Computer Office can be appreciated from the list of programmes given in Table I. Several programmes of a confidential nature have been omitted, but even so some 110 programmes are listed. Programmes 1 to 49 are also omitted since they were written for the Computer at Manchester University which has a different order code than the computer considered here.
Some of the programmes are elementary, such as Reduction of Wind Tunnel Data, but computation is sufficiently repetitive to warrant the use of the Computer in order to speedup the processing of the Data. Most programmes occupy between 10 and 20 minutes running time on the Computer and there are certain programmes relating to stress analysis and Aero-elasticity which run for hours. These latter are the type of problem which make the Computer so important; without it, the time involved in solving these problems would restrict the extent of analysis, with consequent reduction in confidence in the safety of the prototype aeroplane and its crew.
A plausible influence of the Computer is the possibility of reducing the number of design modifications after the prototype aircraft has flown. The increased volume of detailed calculation carried out during the design stage will contribute to the design of a more perfect product obviating the need for many of the modifications which hitherto have been necessary during the flight development of most aircraft.
I am very grateful to the Management of A. V. Roe and Co. Ltd., for their kindness in allowing me to discuss their Organisation so freely. They have my admiration for their foresight and enterprise in installing the Computer at such an early date in the application of these devices.
No.PROGRAMME TITLES 50 Simultaneous Equations/A. 51 Schuerch Wing Analysis. 52 Normal Modes. 53 Matrix Inversion/A,, 54 Flutter Determinant. 55 Balance Calculations. 56 Frame Stressing/A. 57 Flutter Coefficients. 58 Matrixcode 59 Multhopp Subsonic load Grading:- 15 × 2 Collocation Points 60 Honeycomb Sandwich Data/A. 61 Kuchemann Wing Loading. 62 Wave Drag /A. 63 Matrix Inversion/B, 64 Williams Wing Analysis. 65 Multhopp Subsonic Flutter Coeff:-15 × 2 Collocation Points. 66 Corrugation Sandwich Data/A. 67 Variation of Skin Friction with Reynold's No. and Mach No. 68 Wing Lofting Calculations/A. 69 Harmonic Analysis. 70 Weber-Newby Chordwise Loading. 71 Compressible Flow Tables. 72 Supersonic Wind Tunnel Nozzles. 73 Wind Tunnel Data Corrections/A, 74 Multhopp Subsonic Load Grading:- 15x4 Collocation Points 75 Wind Tunnel Data Corrections/B0 76 Weight and Stress Data Scheme. 77 Wind Tunnel Data Corrections/C. 78 Matrix Inversion/C. 79 Matrix Inversion/D. 80 Wave Drag/B. 81 Solution of 1st Order Non-Linear Diff, Equations. 82 Climb Performance/A. 83 Wave Drag/C, 84 Rolling Power 85 Wave Drag/D, 86 Coefficients of Aero, Stability Equations. 87 Wing Divergence, 88 Simultaneous Equations/B. 89 Wind Tunnel Data Corrections/D. 90 Wing Lofting Calculations/B. 91 Matrix x Vector/A. 92 Matrix x Vector/B. 93 Fuel Sloshing. 94 Stress/Strain Curves. 95 Forced Flutter Vibration Analysis. 96 Technical Department Data Sheets. 97 Area Transfer Rule. 98 Heat Transfer Coefficients/A. 99 Climb Performance/B. 100 Integration of Control Surface Derivatives. 101 Multhopp Subsonic Load Grading:-31 x 3 Collocation Points. 102 Rigidity of Sandwich Cores. 103 Calculation of Lift and Moment from CP, Graphs. 104 Anti-Icing Heat Flow. 105 Corrugation Sandwich Data/B. 106 Wind Tunnel Data Corrections/E. 107 Frame Stressing/B. 108 Refuelling System. 109 Open and Closed Loop Frequency Response. 110 Honeycomb Sandwich Data/B. 111 Engine Performance. 112 Flutter Derivatives - Richardson Method. 113 R.A.E. Tunnel Data Reduction. 114 Method of Characteristics for Drag. 115 Etkin-Woodward Wing Loading. 116 Distributed Flange Wing Analysis. 117 Calculation of Pressure Heights. 118 Evaluation of Convolution Integrals. 119 Heat Transfer Coefficients/B. 120 Wind Tunnel Wall Control. 121 Matrix for Wing Pressures. 122 Descent Performance. 123 Calculation of Wing Twist. 124 Solution of Heat Transfer Differential Equations. 125 Response to Atmospheric Turbulence. 126 Wing Stressing by Plate Theory. 127 Expansion of a Determinant as a Polynomial. 128 Design of Engine Intakes. 129 Factored B.M, and Shear Distribution. 130 Single Leg Undercarriage Performance. 131 Cruise Performance. 132 Machine Tool Control. 133 Solution of Heat Transfer Simultaneous Diff. Equations. 134 Engine Inlet Data Reduction. 135 Overall Modes by Branch System. 136 Computer Time Records. 137 Polynomial Quadratic Factors. 138 Supersonic Wing Wave Drag. 139 General Climb and Descent Performance. 140 Whirling Speeds. 141 Take Off Performance. 142 Take Off and Landing Performance. 143 Engine Inlet Data Reduction/B. 144 Two Group Reactor Equations. 145 Large Deflection Analysis of Flat Plates. 146 Weber-Newby Chordwise Loadings/B. 147 Rolling Response with Inertia Coupling. 148 Evaluations of Determinants. 149 Buckling of Parallelogram Plates. 150 Stability under Arbitrary Elevator Motion. 151 Simultaneous Linear Differential Equations. 152 Turbine Disc Stressing. 153 Simultaneous Equations/C. 154 Digital Differential Analyser Simulation. 155 Matrix Inversion Improvement. 156 General Second Order Transfer Function. 157 Thrust Data Reduction. 158 Simultaneous Equations/Da 159 Wave Drag for Optimum Area Distributions.
