**1. Introduction**

The construction sector operates a huge number of test methods. Some of them are used to reveal the truth (in a technical sense) about a material, element, or building, while others are simple conformity assessments of products placed on the market. Single laboratory test may give the impression of an irrelevant issue against the complexity of civil engineering. This makes it hard to consider test result uncertainty in this context. However, taking into account all activities related to construction from the moment of its design through all its life cycle, the tests are present at all stages (Figure 1).

The matrix of building science, shown by Czarnecki and van Gembert [1], its multi-faceted nature and complexity in defining building properties implies a colossal number of research and laboratory test methods used in this field including tests of physical and mechanical properties of building materials, chemical, biological, electromagnetic, electric and electronic tests, acoustic and fire tests of buildings and building elements, tests concerning environmental engineering, tests of radiation of building materials, structural tests of materials and others. On the other hand, there is a centuries-old tradition of following the rules of thumb in this area. For many centuries, it gave quite good results. So what is the point of saying about development of precise and modern laboratory test methods? The reason for the need to improve test methods is the accelerated development of materials and

technologies, which forces the development of new ways of assessing their suitability. Currently, construction engineering ceased to operate only in the meters and reached the nanoscale [2].

**Figure 1.** Presence of tests at all stages of the creation and life of buildings.

The type of test depends on its purpose. Some of them are individual and specially designed, while others are standardized. If we consider the tests methods according to the scheme in Figure 1, most of them, and above all methods used for validation of models or for the purpose of solving scientific issues are rather individual and developed for a specific need. Examples of such individual tests are on macro-scale the experimental and numerical methods used to study the mechanical behavior of objects at a natural scale (e.g., [3,4]), fire tests of walls or doors [5]. Micro-scale tests are among others scientific methods based on advanced material research, such as X-ray diffraction (XRD), energy dispersive spectroscopy (EDS), scanning electron microscopy (SEM) (e.g., [6–8]), and other advanced methods used to understand the mechanisms of phenomena occurring from the macro to the nano-scale. There is also a wide range of individual test and assessment methods used during the service life. Most of them are diagnostic methods such as chemical, ultrasonic, electrical, etc. [9,10] used for building safety inspections. Methods and systems of internal environment testing and sustainability of buildings assessment are rapidly developing (e.g., [11–13]). All test methods in all groups shown in Figure 1 are associated with uncertainty and potential dispersion of the results (the dispersion of some results, especially those from individual methods, is not observed because they are performed only once). The issue of uncertainty and precision of the method is, however, particularly important for laboratory tests used to assess the conformity of construction products. These methods should be standardized and ensure an adequate level of compatibility of test results. In this paper, the laboratory tests are generally divided into the research scientific oriented and routine tests of conformity.

The diagram of this article is shown in Figure 2. The focus is on laboratory tests for conformity assessment because, based on them, decisions are made regarding the possibility of using the material or element in the building. Whenever there is a decision involved, regardless of human activity, there is an uncertainty aspect [14]. Test methods are inseparable from result uncertainty. Realizing the role of uncertainty of the building materials tests results can be easy if we consider that the next, higher assessment level of a complete structure takes into account the sum of partial assessment results, which causes further uncertainty amplification. Thus, the bottom-up procedure is a cluster of sub-models for subsequent assessment models. Result uncertainty may multiply by subsequent uncertainty accumulation at higher assessment levels.

Wherever a decision is related to compliance or non-compliance with specific criteria, this uncertainty involves the risk of committing either a type I error, i.e., rejection of an element which meets the requirements; or a type II error, namely accepting an element which does not meet the

requirements. There is also the risk, however, of committing a broadly defined type III error [15] related to asking a wrong question (using the wrong null hypothesis).

**Figure 2.** Graphic diagram of the article.

Consider the type I risk error in relation to construction materials. This error may incur costs related to the need to create materials or element of a higher performance level than required. This is done in order to avoid erroneous rejection as the result of test uncertainty. The second error type risk is associated with a building's use-phase and ranges from reduced durability and premature loss of properties through construction disaster extreme cases. In view of these analyses, the type III error risk may mean adoption of incorrect assessment criteria, inadequate testing methods, or incorrect result uncertainty estimation, which may result in the case of wrong material assessment. Therefore, each risk is associated with both: cost and environmental impact (waste, materials with higher content of specific ingredients etc.). In an 'era of sustainability' the scientific and research community should strive to minimize such risks at all levels of decisions [16].

Sustainable development in building material engineering is multifaceted and multidimensional [17]. "Sustainable" test method development seems to be part of the progress—such methods help to determine material suitability for specific applications in a credible way, i.e., reducing the risk of threat to life and human health ye<sup>t</sup> avoiding the risk of unnecessary cost generation, excessive use of raw materials, and unnecessary industrial waste generation.

The way uncertainty is taken into account in determining compliance may vary, as outlined in the document of the Joint Committee for Guides in Metrology (JCGM) [18]. When the safe acceptance principle is applied, a test result together with uncertainty must be within the specified limits outlined by permitted tolerance in order to declare a material compliant with the requirements. This material evaluation and material suitability for a specific application in construction involves a typical dilemma: increased cost or increased evaluation safety. This has been discussed in relation to measurements by, e.g., Pendrill [19]. The cost increase encompasses both production costs, including material composition changes in order to be on the 'safe side', as well as costs of precise tests. Evaluation error risk reduction can be achieved through measuring device accuracy improvement and increasing the number of samples.

Experience of laboratory test results indicates that, apart from structural materials for the construction industry, uniform and correct rules for uncertainty estimation for other construction material testing are not available. The uncertainty of measurement paradigm evolution has resulted in divergence in uncertainty treatment and estimation ([20–24]). Today's scientific milieus are inclined to argue that measurement uncertainty is the researcher's statement on the level of knowledge on a measured quantity. What is worth noting is that uncertainty related issues only pertain to measurements as such. Uncertainty, however, is ignored in scientific discussions in relation to test results. There are few guidance documents [25], which recommend treating tests and measurements alike, but this seems unrealistic, especially in the context of testing building materials. The di fference between measurement and test, lies in the fact that in the case of measurements there is a reference value (RV), such as a standard, which is the yardstick to verify trueness. A measurement results should be independent of the measurement method because measurand is physical quantity (like temperature or mass). Uncertainties associated with di fferent measurement methods may be di fferent. A test method is usually specific procedure used to determine the characteristic and there is no reference value for tests. A test quantity definition [26], i.e., a test process model, may be based on several measurements and assessments. A test may o ffer quantitative results or qualitative results. Quantitative results may not always be a function of the measured values. No uncertainty is attributed to qualitative results at the current knowledge stage. Knowledge obtained from the repeatability and reproducibility of experiments is recommended when estimating the uncertainty of test results. Such experiments may be very expensive and labor-intensive and, thus, should not be the responsibility of a laboratory alone, which has utilized a specific test method. Experiments should be the responsibility of organizations, which develop test methods—these bodies are required to validate methods for intended use. The EA-4/16 guidelines indicate that 'laboratories cannot in general be expected to initiate scientific research to assess the uncertainties associated with their measurements and tests' [25].

The ambiguities regarding the basic uncertainty estimation level of test results may result in numerous discrepancies between laboratories. These include model equation discrepancies, considered interactions, estimation of uncertainty components, which stem from result dispersion based on historical data or current results, comparative test results inclusion and di fferent coverage factors. Szewczak and Piekarczuk [27] showed discrepancies in uncertainty estimation and their consequences for final material assessment.

Laboratories di ffer in the level of their research equipment and sta ff competencies. We will consider the ideal situation, i.e., laboratories with proven competences (e.g., accredited), having the same level of equipment and eliminating accidental and subjective errors through detailed analysis and participation in proficiency testing. There will still be discrepancies in results due to a number of factors related to the test method. At the same time, even laboratories with the highest level of competence, based on general guidelines for estimating uncertainty, can receive significantly di fferent values of uncertainty, which then a ffects the assessment of the material tested.

The uncertainties indicated by laboratories may create a wrong impression on the knowledge level of the possible result variability. In some cases, uncertainty is underestimated—this may result in a material obtaining positive assessment results, when such assessment should, in fact, be negative. An overestimated uncertainty used in assessment may, in turn, result in cost increase, which results from the need to improve material properties. It should be noted, however, that both underestimated and overestimated uncertainty values are based on blurred, non-unified prerequisites. Therefore, general principles for considering uncertainty in conformity assessment [18] become irrational if the uncertainty is burdened with even greater uncertainty resulting from the non-uniformity of methods for uncertainty estimation.

The aim of the paper is to show that, when developing test methods, it is not always appropriate to combine the purpose of revealing the truth with the purpose of conformity assessment. Test methods that simulate real conditions may have too many non-controllable parameters that increase the variability of results. This can lead to irrational decisions.

This paper uses simple two independent and contrast test examples, such as bond strength and tensile strength, in order to demonstrate the consequences of using test methods when method suitability for conformity assessment has not been properly verified (i.e., a method has not been properly validated). Uncertainty concept ambiguity in relation to such methods undermines the assessment's rationality. Although in this article the authors considered only examples of mechanical tests, they believe that the conclusions resulting from simple examples are more universal and form the basis for opening a discussion on the test methods development direction, which makes it possible for the methods to be considered 'sustainable'.
