Nordtest bestiller måleusikkerhedsprojekt hos Teknologisk Institut Nordtest, (strategy group on quality and metrology), har i februar 1998 afgivet bestilling til Teknologisk Institut på gennemførelse af den opgave, som er beskrevet neden for. Information om Nordtest: Klik her. Vi inviterer brugerne af internet-stedet www.gum.dk til at gennemlæse projektbeskrivelsen. Afføder dette interesse eller kommentarer, kan der rettes henvendelse til Sven Nytoft Rasmussen, e-mail: snr@dti.dk, telefon +45 43 50 44 42, telefax +45 43 50 72 73. (Projektsproget for denne opgave er engelsk): Tools for the test laboratory to implement measuring uncertainty budgets. by Sven Nytoft Rasmussen, Centre Manager, Ph. D, Danish Technological Institute. Project summary: Uncertainty budgets are expected to become ever more relevant to Nordic testing laboratories. Today's tools for managing and computing such budgets do not allow for cost effective handling of uncertainty. The project shall study novel tools and techniques. A Nordic project group equally open to European participation shall be set up to discuss the results. Dessimination of results as per Internet publication as well as written report in Nordtest publication series. Our partners would benefit from sharing our present know-how and access to the latest software tools.   Background and motivation Much attention has recently been paid to the question of including in official test reports a statement of the measurement uncertainty associated with a test result. Starting from the late 1980'es the requirement of measurement uncertainty laid down in EN 45001:1989 was rigorously interpreted by Nordic accreditors working with calibration laboratories. In the case of testing laboratories much work remains to be done in the Nordic countries. Only recently has a strict interpretation of the requirement of measurement uncertainty been imposed as an accreditation criterion for testing laboratories. The EAL has published a guideline on the subject (EAL-G23 1996). Recent announcements of Nordic accreditors have stressed that the application of modern uncertainty budgeting principles must now find its way into testing laboratories. A process is sought to swiftly introduce measurement uncertainty according to the International Guide to the Expression of Uncertainty in Measurement (GUM-method). Measurement Uncertainties are a requirement – according to EN 45001:1989 "where they are relevant". The recently introduced interpretation, stressed by the forthcoming introduction of a new edition of EN 45001, of the term "where relevant" is such, that measurement uncertainties are always relevant and should be applied, whenever an accredited testing laboratory issues quantitative results.   Internal obstacles, proposed tools: The GUM-method introduces a way of thinking and requires a set of mathematical methods not easily understood by typical testing laboratories. In many laboratories measurering uncertainties have traditionally been neglected all together. Elsewhere i.a. in most analytical laboratories measuring uncertainty has been handled using traditional statistical methods. Laboratory personnel requires additional training before they can work according to the new practices. Measuring uncertainty budgets are by some laboratories looked upon as an additional cost which cannot be justified by added income. Thus, fully implementing GUM method principles may pose a demanding task for laboratory management. As compared to a calibration laboratory, the situation typically is more difficult for a test laboratory. The test laboratory handles several measuring disciplines often using complex measuring equipment. In most cases, a test result is derived from several metrological signals thus calling for a measuring equation with related sensitivity coefficients. Since tests may be time consuming to carry out, the test laboratory is often faced with limited repeatability studies, thus resulting in a wide variety in the degrees of freedom which can be associated with the various components of measuring uncertainty. This in turn leads to more elaborate considereations in determining the coverage factor. In a testing laboratory, any one piece of measuring equipment is typically applied in several testing methods, while it is typical for most calibration laboratories, that calibration equipment is reserved for just a single application. Also symptomatic for the situation in a testing laboratory is, that the use of measuring signals and measuring equipment may vary due to the many practical concerns in testing, where methodologies must often be slightly varied from test to test. Because of high demands on flexibility a testing laboratory offen needs measuring uncertainty budgets to be devised ad hoc. Budgets do not remain universally valid once worked out. Among the tools to facilitate the implementation of uncertainty budgets are training courses. In this part of the study, only little emphasis will be placed on evaluating the training courses avaliable for testing. A major part of the study should focus on best practices for the handling of data and the process of continuously updating uncertainty budgets. It should be examined which software solutions are appropriate for the needs of Nordic testing laboratories. Off-the-shelf software like e.g. Uncertainty Analyser by ISG and Expression of Uncertainty by the German start-up company Metrodata Gmbh should be considered. By the time of evaluation, also Grachanens/Compaq's Uncertainty Calculator is expected to fully cover the GUM method and thus should enter into our study. The latter software tool is duly supplemented by another package labeled Conversion/Expression Buddy by J. Presley. The software should be evaluated and tried out for typical uncertainty problems collected among the participating laboratories. Suppliers of such software claim that using their software, the test technician need not concentrate on any mathematical manipulation. He only needs to combine his knowledge of the sources of uncertainty in his special field of testing with the mathematics built into the software. The software thus enables testing laboratories to administer valid uncertainty budgets using the skills and expertise already at hand. The project should discuss the validity of such claims. The features and limitations of the software applications should be examined. Also the possible role of such software in the overall Laboratory Information Management System (LIMS) should be studied. In this consideration it is suggested that the Nordic testing laboratories may put the principles of the socalled PUMA method to good use. This method (Procedure for Uncertainty Management ref. ISO/DTR 14253-2 – ISO/TC 213 N67) is an iterative method which may significantly help laboratories to arrive at adequate uncertainty budgets with a minimum of labour. Also the problem of validation of standard software as well as custom built software solutions for the estimation of measurement uncertainty should be discussed in the project. For this a few well worked sets of testdata should be made available. A proposal for the desired specification for integrated software should be given. Based on the combined competencies of the project partners a balanced view could be presented guiding future initiatives by software developers and laboratory managers. Finally the project should develop recommendations for documentation of uncertainty budgets. Reporting formats suitable for the types of uncertainty budgets most often used by testing laboratories should be laid down to aid discussions with clients and accreditors – as well as future communication within the laboratory.