The continuous progress made by data processing technology and the dramatic decrease in the cost of its hardware allow physicists to imagine evermore sophisticated machines to help break through physics barriers. As in many industrial processes, ex­ perimental physics facilities are today unthinkable without a fleet of compu­ ters to ensure their smooth running. Par­ ticle accelerators for high energy phy­ sics and telescopes for astronomy illus­ trate research areas where many recent discoveries are strongly indebted to computer controls.

A 5 day workshop was held in November 1988 at Los Alamos National Laboratory to address the viability of providing a toolkit optimized for building accelerator control systems. The workshop arose from work started independently at Los Alamos and CERN. This paper presents the discussion and the results of the meeting.

The CERN Technical Board for Process Controls and Accelerator Electronics has addressed the issue of software engineering in order to improve the development of software. This study has been extended within the frame of the interdivisional group on Experimental Physics Control Systems of the European Physical Society. The results of this study are reported briefly in this paper.

This paper presents the author's view on how application software could be structured into generic packages which, at the cost of limited programming only, could be tailored to suit the needs of daily operation and use of accelerators. The ideas dealt with in this paper are not claimed to be new or original. They result from observation of efforts at CERN [IEEE Trans. Nucl. Sci. NS-32 (1985) 2092; Proc. Europhys. Conf. on Control Systems for Experimental Physics, Villars sur Ollon, Switzerland, 1987, CERN Yellow Report, to be published], other laboratories [Nucl. Instr. and Meth. A293 (1990) 321] and industry to rationalize application software in the quest for higher efficiency and better quality, and under the pressure of ever increasing demand and shrinking resources. The selection of publications to which reference is made illustrates these activities.

Throughout their history the controls of the CERN PS accelerator complex have followed the general trend towards distribution of its processing power. The current system that contributed to CERN's recent achievement in the field of particle physics was designed in 1978, and involves a network of 20 minicomputers and 150 microprocessors. At that time it was a state of the art distributed system. The advent of a new accelerator project, the preinjector for the Large Electron Positron machine, gave the opportunity for an in depth review, in particular of the structure of its application software, in the light of experience, new operational requirements and advance in technology. The salient feature of the LPI controls is that it goes one step further towards distributing controls power whilst preserving full compatibility with the existing system. Most of the activities that are currently executed by front end minicomputers will be entirely delegated to modern more powerful microcomputers. The application software is now being restructured as is described in this paper. In conclusion the status of the project is given and some figures relating to the effort invested in this new system.

The CERN PS accelerators have evolved into one of the world's most sophisticated high energy physics facility. The variety of beams and their high repetition rate means that a most sophisticated controls system is required. This reflects on the application software. At the time of the completion of the new control system, nearly 1000 programs, amounting to around 450 000 lines of code, have been developed at the cost of approximately 120 man-years. The span of this software ranges from real-time application programs to special purpose development and management tools.

Current PS controls application software has a strong hierarchical structure of software modules which translate user-friendly commands into the intricacies of hardware devices. Starting from the most elementary hardware level, these are: (i) the “interface module” (IM), which hides the various CAMAC commands so as to provide a standard access to each type of CAMAC module; (ii) the “equipment module” (EM) presents a simple standard software interface of each process equipment. There is one EM for each type of equipment and all process equipment is accessed through EMs via IMs; (iii) the “composite variable module” (CVM) provides control of abstract beam variables. It involves control of several and possibly different kinds of equipment. Setting a CVM results in appropriate setting of all relevant equipment through calls to their EMs.

The CERN PS accelerator complex has been progressively converted to tull computer controls without interrupting its tull-time operation (more than 6000 hours per year with on average not more than 1% of the total down-time due to controls). The application software amounts to 120 man-years and 450'000 instructions; it compares with other large software projects, also outside the accelerator world: e.g. Skylab's ground support software.1 This paper outlines the application software structure which takes into account technical requirements and constraitnts (resulting from the complexity of the process and its operation) and economical and managerial ones. It presents the engineering and management techniques used to promote implementation, testing and commissioning within budget, manpower and time constraints and concludes with experience gained.

in the future. Because of the well-known environmental impacts of synthetic fuels plants, an extensive permit process has been devised by local, state and federal regulatory agencies covering the construction, operation and maintenance of the proposed Ethanol Refinery. This permit process, in the absence of any unusual circumstances, serves to eliminate the need for an Environmental Impact Statement. Socioeconomic effects are minimal due to the small number of employees (49) required to operate and maintain the Ethanol Refinery. However, the use of grain as feedstock for producing the ethanol provides a year-round market for locally-grown crops. The economic benefits of this are significant to the local farmers and businessmen since it will substantially reduce the seasonal ups and downs of the local economy.

This paper presents the economics of the application software developed for a large process control system. The structure of the application software will be discussed in both a technical and economical perspective, and assessed with Barry Boehm's development cost estimating formalism. The results will be compared with the in-house estimates and the actual costs.

The Antiproton Accumulator (AA) is a technically advanced, complicated and unusual facility. Its controls had to be built in a short time by a very small permanent staff calling on part-time resources from other parts of CERN. As the AA will eventually be integrated into the PS accelerator complex, its process interface, process computer, systems software and lower level applications software have been made to the new PS standards. For the efficiency of commissioning and first operation the present versions of high level application programs have been written in NODAL by the equipments specialists using a programming philosophy and a new "Touch Terminal" from the SPS. In addition, two desk top computers with a separate GPIB are used for the stochastic cooling system. The paper outlines the pragmatic approach in composing the controls for a complicated machine using the procedures and components available at CERN. The experience gained after an early and successful start-up is reported.

The new CPS controls system was successfully put into operation on 19 November, 1980, when the 800 MeV Booster, after switching over to the new system, was started up for particle physics. A subset of the new system had already been installed on the Antiproton Accumulator in Spring 1980 and has reliably served several months of engineering run. The Booster machine and control of the CPS cycle gene...

Computer assisted controls at the 28 GeV PS made their entry in 1967 and today around 80% of the processes are included in various styles. Beam intensity has since increased two orders of magnitude and interleaved cycles of different beam properties are now serving SPS, ISR and the 28 GeV experimental area. This came about by substantial additions to the accelerator equipment, the main one being the Booster and Linac. Plans up to the end of 1980 include: addition of the Antiproton Accumulation Ring, acceleration of antiprotons in the CPS, the concomitant beam transfer and switching, and multibatch filling of the SPS, requiring cycles times down to 0.65 sec. The improvement programme for controls aims to alleviate the operational and maintenance problems ensuing from this explosive expansion and to create a framework for further growth.