M experiments

M experiments

World Abstracts on Microelectronics and Reliability A body of Engineering Change Proposals (ECPs) was used as source data. They had all been formally ...

257KB Sizes 2 Downloads 74 Views

World Abstracts on Microelectronics and Reliability A body of Engineering Change Proposals (ECPs) was used as source data. They had all been formally presented to the customer by Hughes Aircraft Company as a result of continuing studies of the performance of a fire control system and associated test equipment and had all resulted in change in contract cost to the customer. The costs referred to hereafter will be limited to these contract costs. The ECPs were broken down into samples according to subsystem in which the change was to be located, purpose of the change (improvement in performance, reliability, safety, etc.) or combination of location and purpose of the change. For the sample and each subsample, relationships were sought between costs (and breakdowns thereof) and parts and/or Integrated Logistics Support (ILS) elements (and breakdowns thereof). Existing computer routines were used to perform correlation, regression and evaluative computations. For the total sample and most of the subsamples, one to three equations were found which could predict some cost element based on either one or two hardware or logistics element(s). For each prediction, confidence intervals are provided at the 80, 90 and 95% levels. Results were compared to actual costs. The models thus provided can be used with a set of simple instructions usually requiring only one value of one variable to be input. They can be used to determine if a contemplated ECP is economically feasible or to choose the least costly of several alternative modes of system change. A support system life cycle cost model. CYR1L BROOK and RAMESH BARASIA. Proc. IEEE Reliab. Maintainab. Syrup. Philadelphia, Jan. 18 20, 1977. p. 297. Bell-Northern Research has recently completed a feasibility study, under contract to the Canadian Department of National Defence, to establish management guidelines for the evaluation of Automatic Test Systems (ATS) requirements. This paper presents an overview of these guidelines, and discusses a case study which was performed to demonstrate the applicability of the recommended procedures. A graph is presented which shows the relationship between the prime system availability, and the support system Life Cycle Costs. A reliability control tool during manufacturing. HERMAN D. RUE. Proc. IEEE Reliab. Maintainab. Symp., Philadelphia, Jan. 18-20, 1977. p. 275. Product reliability events normally recorded in manufacturing failure data collection systems can be combined as a series of factors whose sum offers a quantitative production "figure of merit". The methodology permits monitoring of reliability trends during manufacturing by comparison of periodic figure-ofmerit analyses. Although greatest benefits can be obtained during long, uninterrupted product manufacturing; the figure-of-merit methology is found to be a valuable product reliability control "tool" even for short-runs.

Failure modes and effects analysis by matrix method. GILBERT L. BARBOUR.Proc. IEEE Reliab. Maintainab. Symp., Philadelphia, Jan. 18-20, 1977. p. 144, A matrix method for analyzing failure modes and effects has been developed that supplements the standard narrative-tabular and fault tree methods. The new method uses condensed graphic displays of vertical and horizontal lines so that failure effects on the highest system level can be traced down through subsystem and lower equipment levels to the contributing piece part failure modes. In addition to this traceability feature, this approach provides more effective accountability of such items as piece parts, interface connector pins, circuit solder joints, and wiring. This capability becomes invaluable in identifying single point failure modes, that is, single failures that can cause catastrophic loss or major degradation of system performance. The

647

matrix method of analysis can also be used for safety hazard analyses, for determining test and telemetry failure detectability, and for preparing maintenance malfunction location charts.

Using pattern recognition in product assurance R. A. HUGHES, M. A. FISCHLER and H. E. RAUCH. Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia, Jan. 18 20, 1977. p. 101. This paper shows how pattern recognition techniques can lead to the solution of problems in product assurance which did not appear to be open to solution using only conventional statistical or mathematical approaches. The techniques described here include conventional statistical analysis as well as discrimination or classification (Fisher Linear Discriminant), hypothesis generation (minimum spanning tree clustering and centroid clustering), data display (two dimensional projection), and data base handling and interactive graphics. This paper presents an overview of methods and experience gained in using the above combination of techniques for exploratory data analysis, decision making, and monitoring of manufacturing processes in problems involving product assurance and quality control. Solving reliability models of nuclear systems. LEONARD R. DOYON. Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia, Jan. 18-20, 1977. p. 322. In this investigative study the author derives a Markovian state-transition diagram for the reliability/availability/maintanability model of a typical coolant water system for a nuclear-reactor powergeneration plant. This system has a considerable amount of redundancy in equipment. The times-between-failure obey the exponential probability law for electronic equipments and the normal probability law for mechanical apparatus. The times-to-repair obey the lognormal probability law for electronic equipments and the normal probability law for mechanical equipments. Fixed-head disk memory unit for high reliability applications. I. ISOZAKI,H. MATSUMOrOR, O. MIYAZAKI, M. ITO, S. NAKAMURA,S. YAMAMURA,S. KAGEYAMA,S. MITSUASHI and K. NAKAMURA. NEC Res. Dev. 44, 57 (January 1977). A fixed-head disk memory unit with high reliability and long life was developed for file use in computers and electronic switching systems. This unit has a maximum capacity of 1.6 million bytes and is expected to have a 22-year life without any overhaul. It is designed to function at a high transfer rate and short access time with high reliability under a wide range of ambient conditions. To obtain the said reliability and long life, the following items are taken into consideration. The unit (1) has a mechanism for replacing grease in bearings, (2) uses materials and finishings in rough conformity to UL (Underwriters Laboratories) Standards and Tropicalization Treatment, (3) adopts monolithic heads, (4) employs a simplified mechanism and high reliability electronic parts and (5) is provided with error check circuits. Though the unit has all these high qualities, it is small in size and low in cost. This disk unit is the most appropriate for data file and program store use in electronic switching systems, control computers and on-line computers which require high reliability. Statistical design of R/M experiments. CHESTER H. MCCALL JR. Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia, Jan. 18-20, 1977. p. 194. In this paper, the basic concepts of experimental design, from a statistical viewpoint, will be summarized. Primary emphasis is on terminology and philosophy. Following this introduction, the concepts will be related directly to potential requirements in R/M testing. No examples are worked out in detail,

648

World Abstracts on Microelectronics and Reliability

but rather are utilized to reinforce the introductory experimental design concepts.

Reliability growth cruves for one shot devices. AUSTIN J, BONIS. Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia. Jan. 18-20, 1977. p. 181. Here is a simple and practical procedure for plotting reliability growth curves for one shot devices. These are tested in stages of development. The results of each stage of testing are used to improve the item prior to testing in the next stage. A steady improvement effort is assumed. The model presented (The Modified Exponential Growth Curve) can be used before testing to plan the test program. After testing, the model will permit measuring and predicting reliability. The model should lead to a valuable planning system for new programs. This is a "how to" paper. Theory has been put in references and appendices. Sample tables are included to facilitate computations for "demonstrating" reliability and estimating the number of trials to be made at each stage. The method is illustrated with a worked out example. Only simple algebra is used. The "three point" method is sufficient for the plotting. MTBF and yield prediction system. ANTONIN WILD. Proc. IEEE Reliab. Maintainab. Syrup, Philadelphia, Jan. 18-20, 1977. p. 170. This paper describes the structure and operation of a computer system for predicting MTBF and production yield of units. The characteristic features of this system are its user oriented design and the backing of failure rate and yield data by tracking both production testing and repairs of equipment in the field.

PMR, NDE, design practices; present and future. NORMAN G. DENNIS. Proc. IEEE Reliab. Maintainab. Symp., Philadelphia, Jan. 18-20, 1977. p. 164. Various kinds of probabilities and types of information are used in the Prediction of Mechanical Reliability (PMR). Confidence in using an established Design Practice in a new product seems to be a Logical probability, as defined in the paper. Non Destructive Evaluation (NDE) increases a designei"s span of control over his manufactured product; it makes PMR more precise via closer estimation of probability; it improves reliability when inverior units are rejected. Thus NDE fits into and improves traditional design practices, manufacturing controls and inspection techniques. Four classes of useful information from NDE for PMR are identified: voids and cracks; gross correlations; organized comparisons of one serial number with others: and finally, sophisticated information derivable from the surfaces of actual components, including relations between grain size and Yield and Ultimate stress.

MTBF prediction by programmable calculator. DOUGLAS W. BOND. Proc. IEEE Reliab. Maintainab Syrup., Philadelphia, Jan. 18-20, 1977. p 264. Among the applications of desk programmable calculators is the solution of the failure rate equations provided by MIL-HDBK-217B for discrete electronic parts. This paper examines the potential for interactive communication with these versatile machines to simultaneously reduce the drudgery and increase the utility of the failure rate prediction process.

Repair limit suspension policies for a two-unit standby redundant system with two phase repairs. S. OSAKI and K. OKUMOTO. Microelectron. Reliab. 16, 41 (1977). This paper proposes a repair limit suspension model for a two-unit standby redundant system with two phase repairs, where a preventive maintenance for a unit under repair is made. A Markov renewal process is applied in analysis of the stochastic behavior of the system. The optimum repair

limit suspension policies maximizing the system availability or minimizing the expected cost per unit time it~ the steady-state are respectively discussed using the above results. Numerical examples are finally presented.

RIW experiences at ECOM. RONALDA. MLINARCInK.Proc. IEEE Reliab. Maintainab, Symp., Philadelphia, Jan. 18-20, 1977. p. 257. The Reliability Improvement Warranty (RIW) concept is currently being utilized by the Army on a trial basis to improve field reliability and thereby reduce support costs. The principal feature of the RIW approach is that all repair (except minor work such as replacing lamps, fuses, etc) is performed at the contractor's facility, The US Army Electonics Command (ECOM) currently has two RIW programs in progress (CONUS NAV and the Absolute Altimeter), and several others are under consideration. Each trial procurement has certain unique characteristics which require that the terms and conditions of the RIW clause are tailored so that the rewards and risks to both the contractor and the Government are acceptable. This paper describes the RIW requirements for each program and examines the bids submitted by the offerors in relation to the warranty. The development of each program is presented with emphasis on results, experience, and lessons learned thus far. Quantitative models used in the R1W decision process. R. K. GATES, R. S. BICKNEL and JorfN E. BORTZ. Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia, Jan. 18- 20, 1977. p. 229. Recognition is being given to the potential value of the Reliability Improvement Warranty (RIW) provision in acquiring reliable military equipment and controlling life cycle cost. The decisions that must be made in conjunction with RIW implementation, both by the government and the contractor, are identified and quantitative models for the decision maker are presented. A case study is included to illustrate the decision processes leading up to contract award.

Using procurement incentives for technological innovation. CHARLES HULICK and JOSEPH G. BERKE. Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia, Jan. 18-20, 1977. p. 78. The Experimental Technology Incentives Program is engaged in conducting procurement experiments with participating agencies to test the hypothesis that government procurement can through the use of procurement incentives induce private sector technological innovation. Experiments are currently underway at the Federal, State and local levels with participating agencies. The result of the ETIP work will be a summation showing which combination of individual agency characteristics and procurement incentives are most effective in bringing forth technological innovation.

Reliability proving for commercial products. DEAN E. BEACHLER and W. A. CHAPMAN.Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia, Jan. 18-20, 1977. p. 89. Product reliability considerations permeate all phases of development and production. This paper describes the reliability proving process used at NCR and illustrates usage of the Duane reliability growth approach.

A model for an estimation of the product warranty return rate. KYUNG Woo YUN and FRANK E. KALIVODA. Proc. IEEE Reliab. Maintainab. Syrup., Philadelphia, Jan. 18 20, 1977. p. 31. Effective cost control requires an accurate understanding of product failure costs. Expensively gathered failure data often, however, cannot be utilized effectively because of possible inaccuracies. This paper develops a simple model for estimating realistic warranty return levels corrected for human errors in the determination of product