Investigation of edge strength dependent on different types of edge processing Jonas KLEUDERLEIN *, Frank ENSSLEN a, Jens SCHNEIDER b * Technische Universität Darmstadt, Institut für Werkstoffe und Mechanik im Bauwesen Franziska-Braun-Straße 3, 64287 Darmstadt, Germany kleuderlein@iwmb.tu-darmstadt.de a b Semcoglas Holding GmbH, Langebrügger Strasse 10, 26655 Westerstede f.ensslen@semcoglas.de Technische Universität Darmstadt, Institut für Werkstoffe und Mechanik im Bauwesen Abstract For the design of glazing, the edge strength of annealed float glass becomes more and more important. In current standards the edge quality is defined by visual and geometric characteristics only. The strength of the glass edge, depending on the type of processing, however, is considered insufficiently in standards. In cooperation with the working group edge strength at Fachverband Konstruktiver Glasbau e.v. (FKG) different types of edge processing were investigated experimentally, and evaluated statistically. Thereby, several processing factors and glass thicknesses were taken into account. For comparison to the definition in the current standards the edge quality was documented visually. The statistical analysis of the experiments gives a first estimate for the correlation between edge strength and the edge processing, considering the variation of production factors and glass thicknesses. Keywords: edge strength, edge processing, statistical evaluation, microcracks 1 Introduction Temperature differences within glass panes cause thermally induced stresses. For usual glass dimensions (thickness << length) and solar loads this can lead to a state of plane stress and depending on the temperature profile to tensile stresses parallel to the glass edge. Furthermore, the use of triple IGU means increased weights and edge offsets which require a higher strength of the glass edge. Thus, the edge strength is indispensable for the evaluation of thermal induced stresses in current facade constructions (cf.[1]). 1
In the current standards the edge quality is documented differently and insufficiently with regard to the edge strength. In DIN 1249-11 [2] only the edge shape in correlation to the types of edge processing is defined. For the thermal stress calculation model of the draft european standard (preen thstr, [3]) allowable temperature differences are suggested which correspond to edge strengths for different types of edge finishing. In DIN 18008 [4], the characteristic edge strength of annealed float glass is defined as 80 % of the surface strength regardless of the edge quality. The edge quality of annealed float glass was investigated several times in other studies (cf. [5], [6], [7]) but in each of them only one or two manufacturers were investigated. This study will provide results of the influence of the manufacturing process on the glass strength concidering six manufacturers, three different types of edge finishings and three different glass thicknesses. The following results were obtained in cooperation with the working group edge strength of Fachverband Konstruktiver Glasbau e.v. (FKG). 2 Experimental procedure 2.1 Scope of testing Three types of edge quality as-cut, arrised and ground according to DIN 1249-11 [2] were analysed (see Figure 2). Therefore, tests were done on nominal glass thicknesses of 4 mm, 6 mm and 8 mm. The test program includes a total of 33 series, each with 30 test specimens. The specimens have the dimensions of 1100 mm x 125 mm and were produced by a total of six manufacturers (Table 1). The manufacturer paid special attention to the fact, that the processing was done with the usual settings of production parameters and all specimens were stored at least for 14 days (crack healing) before transportation. In order to minimize possible damaging, the specimens were stored and transported vertically with the testing edge upwards. 2 Type of processing 4 mm 6 mm 8 mm as-cut M1; M2; M5 M3; M4; M6 M1; M2; M3; M4; M5 a arrised M3; M4; M5 M1; M2; M6 M1; M2 a ; M3; M4; M5 ground M2; M3; M6 M1; M4; M5 M1; M2; M3 a ; M4; M5 Quantity of specimens 270 270 540 Table 1: scope of testing and manufacturers ( a double test series for reference tests), M = manufacturer 2.2 Testing method The specimens were tested in-plane using a four-point bending test until breakage. The specimens were supported by synthetic rollers with a distance of 1090 mm. To prevent
tilting, synthetic-coated stabilizer were arranged. The tests were carried out with a load span of 200 mm and a stress rate corresponding to the glass thicknesses of 2 MPa/s in accordance to DIN EN 1288-3 [8] (Figure 1). Only those tests were considered in which the fracture origin was within the load introduction. The region of constant stresses of the testing corresponds to the plane stress state under solar loads with the assumption of a constant temperature gradient over the glass thickness. load application test stabili support roller Figure 1: Four-point bending test The experiments were carried out at the Materialprüfanstalt Darmstadt and the Institut für Baukonstruktion at the Technische Universität Dresden. To ensure the comparability of results, preliminary tests on identical test equipment at both locations were done. Detailed information to the scope of testing and the testing method can be found in [1]. 3 Results and discussion 3.1 Manufacturing parameters The process parameters for each working step were documented by the manufacturers. The manufacturing protocols show that all process steps are machine made. However, the manufacturers use almost individual process parameters. Thus, manufacturers use cutting machines of different companies. In addition, different cutting pressure, different cutting speed (60 m/min to 160 m/min), cutting wheels with different angles (140 to 156 ) and different cutting fluids (e.g. ACECUT 5503, 5476 or 5250) were used, even for the same glass thickness. The edge quality arrised is produced differently by the manufacturers. Half of the manufacturers use edging machines with cup wheels to arrise the glass edges, the others use cross belt edging machines with manual feed. The speed for automatic feed varies considerably among different manufacturers. The belt speed for cross belt machines also varies between 1400 m/min and 1800 m/min and for edging machine with cup wheels the rotation speed varies between 2400 rpm and 2900 rpm. It is noteable that by some 3
manufacturers ground edges with blank spots for the quality arrised edge were delivered (see Figure 3). All manufacturers use a cup wheel edging machine for grinding. Again, manufacturers use machines of different companies. The feed speed differs between 2 m/min and 5 m/min and the grinding allowance between 1.0 mm to 2.0 mm per side. The suppliers of the cup grinding wheels differ as well. Figure 2: Investigated types of edge finishing: as-cut (left), arrised (middle) and ground (right) according to [2] Unfortunaltely, some process parameter based on machine protocols are not compareable and thus the documentation is sometimes incomplete (e.g. grinding pressure). Both the grinding and the arrising process use different grain sizes in varying order. The diamond grit sizes vary in the range of 54 µm to 325 µm as well. In addition to the process parameters, the optical quality of the glass edge was documented. As expected from the manufacturer's protocols, a considerable band width in the optical quality can be found (see Figure 3). Figure 3: Examples of edge finishing by different manufacturers for the quality as-cut (first line), arrised (second line) and ground (third line) according to DIN 1249-11 [2]. Top view on the edge surface. 4
It should be noted that so far the influence of manufacturing parameters on the edge strength was not investigated sufficiently. Usually, in practice the optimization goal for the parameters is an easy opening glass cut and the visual quality of the glass edge. But these goals do not necessarily correlate with the strength of the glass edge, as is shown below. 3.2 Experimental results and discussion 3.2.1 Fracture origin For the specimens tested at the MPA Darmstadt the location of fracture origins were evaluated (see Figure 4). The positions at the cutting border, on the edge surface and at the opposite border were differentiated. Thereby for arrised and ground edges, the term "border" refer to the border to the glass surface. According to the type of edge processing, differences in the distribution of fracture origin can be identified (see Figure 4). Quantity of fracture origin 200 180 160 140 120 100 80 60 40 20 0 8 182 72 47 59 55 as-cut arrised ground cutting border opposite border Figure 4: Distribution of the fracture origins in correlation to the quality of the edge finishing and the definition of the cutting border, the edge surface and the opposite border (right) 105 18 o.b. e.s. c.b. The distribution of fracture origins suggest that the damages in the form of microcracks caused by the cutting process (Figure 5) are eliminated or at least significantly reduced by arrising or grinding the edge. These results are consistent with other studies (cf.[7]). However, new damages are introduced by the arrising or grinding process that can be less favorable in terms of strength as the damages introduced by the cutting process. 5
Figure 5: Edge quality as-cut : Top view on the glass surface (magnified 200 times) for the cutting side with microcracks (left) and on the opposite side (right) 3.2.2 Statistical comparison of the edge strength The investigation is based on 33 test series with 830 valid failures. In addition to the mean value and the coefficient of variation, the 5 % fractile at 95 % confidence level assuming a two-parametric Weibull distribution were determined for each series. The goodness of fit for the chosen distribution in the context of edge strength is shown in [6] and was also confirmed in this study by comparing single series with normal and lognormal distribution. Weibull and lognormal distribution do not show significant differences for the given data sets with relatively low coefficients of variation. In Figure 6, the inverse function of the Weibull distribution over the bending strength with the corresponding linear regression for the individual series and for all series combined is shown. The gradient of the linear regression represents the scattering of the series. The coefficient of determination R² is used to evaluate the prediction of the approximated regression in a first step. For estimation of characteristic design values R² close to 1.0 is desirable. Figure 6 is representative for the distribution of data sets within the group of comparable edge quality and glass thickness. The following can be observed: - The individual series have a low scattering (coefficients of variation v < 0.15). The coefficient of variation for bending strength for glass surfaces of float glass is usually v = 0.20 to v = 0.30 (see e.g. [9]). - If the data sets of several series are combined in a distribution, the quality of the prediction decreases significantly. - Series with the same edge finishing show a similar gradient of the linear regression, but with respect to the strength they are on different levels. This suggests that the machining process achieves reproducible strengths but different production parameters lead to different strength levels. 6
2,5 1,5 0,5 R² = 0,9726 R² = 0,9732 R² = 0,909 R² = 0,7951-0,5 Inv (F) -1,5-2,5-3,5 R² = 0,9514 R² = 0,9415-4,5-5,5 3,5 3,7 3,9 4,1 4,3 4,5 LN(σ) [LN(MPa)] Figure 6: Probability plot of the Weibull distribution for the individual series of 8 mm arrised samples and for all series combined (blank circles) To investigate the influence of the edge machining on the edge strength for each series in detail, the mean value, the 5 % fractile, the mean value of the fractiles and the corresponding standard deviation are compared (Figure 7). The three diagrams show the results separately for a glass thickness of 4 mm, 6 mm and 8 mm. The following can be observed: - The scattering of the fractiles resulting from the different processing superimpose the grouping of the series by the processing mode as defined in the standards. This means that the influence of different production parameters is at least in the same range as a possible increase in strength by further process steps. - The mean values of the fractiles show no clear trend, which points out an improvement of the edge strength by arising or grinding. - An increase in the edge strength can be achieved. However, the strength may also be reduced by the further process after cutting. Again, the production parameters have obviously a significant influence (see also [5]). The above statements apply both when considering the mean values of the series as well as in consideration of the fractiles. 7
failure stress [MPa] failure stress [MPa] failure stress [MPa] Figure 7: Failure stresses, mean value and 5 % fractile for each data set for the glass thickness of 4 mm (above), 6 mm (middle) and 8 mm (bottom) glass thickness. The mean value of 5 % fractiles of all series within each group (not to be confused with the 5 % fractile of all merged data of each group) with the corresponding standard deviation are indicated as continuous and dushed lines. 8 100 90 80 70 60 50 40 30 100 90 80 70 60 50 40 30 100 90 80 70 60 50 40 30 as-cut arrised ground as-cut arrised ground as-cut failure stress arrised ground mean value 5% fractile mean value of fractiles
Overall only two out of 33 fractiles are below the suggested characteristic edge strength of 36.0 MPa according to DIN 18008 [4] (45 MPa 0,8 = 36 MPa) (see Figure 8). The minimum value of the 5 % fractiles is 34.23 MPa, the maximum value of the 5 % fractiles is 64.84 MPa. characteristic stress [MPa] 70 60 50 40 30 20 10 0 as-cut 4 as-cut 4 as-cut 4 ground 4 ground 4 ground 4 arrised 4 arrised 4 arrised 4 as-cut 6 as-cut 6 as-cut 6 ground 6 ground 6 ground 6 arrised 6 arrised 6 arrised 6 as-cut 8 as-cut 8 as-cut 8 as-cut 8 as-cut 8 ground 8 ground 8 ground 8 ground 8 ground 8 arrised 8 arrised 8 arrised 8 arrised 8 arrised 8 Figure 8: 5 % fractiles for all series in relation to the characteristic edge strength according to DIN 18008 [4] 4 Conclusion and summery test series f k, float, edge according to DIN 18008 The band width of possible manufacturing of the glass edge for the investigated edge qualities as-cut, arrised and ground according to DIN 1249-11 [2] is shown by the scope of testing with six participating manufacturers. The optical quality of the edges differs significantly in correlation to the manufacturers. The edge strength is not correlating with the optical quality of the glass edge. Especially for the arrised edge, different finishing processes (cross belt or cup wheels) and their parameters can obviously have a significant influence on the edge strength. Furthermore it is shown, that an increase in the edge strength could be achieved by arrising or grinding. However, the strength may also be reduced by this edge finishing after cutting. In general, series with the same edge finishing show a similar scattering, but with respect to the strength they are on different levels for different manufacturers. This suggests that the machining process achieves reproducible strengths, but different production parameters lead to different strength levels. On the basis of the presented results the authors consider the proposed characteristic edge strength of 36 MPa according to DIN 18008 to be appropriate without differentiation of the edge finishing. In order to achieve higher 9
characteristic values of the edge strength, additional effort in influence of manufacturing and assembling is required. This has to be approved by further research. It should be noted that the strength is only one characteristic of a glass edge. Arrised and ground edges offer further advantages regarding to the risk of injury or the glass breakage during transportation and installation. For a better unterstanding of the correlation of processing and the edge strength further parameter studies are required. For the evaluation of thermal breakage the authors still see considerable need for research regarding a design basis for thermal loads. Effects of scaling and load duration will influence the results shown and need to be investigated in future more closely. 5 Acknowledgement The authors gratefully acknowledge the support of the Fachverband Konstruktiver Glasbau e.v. (FKG) and in particular the members of the working group edge strength who made it possible for this project to be realized by delivering and documenting the test specimens. Special thanks to the companies Aachener Chemische Werke and Bohle AG for the substantial know-how of the edge processing. 6 References [1] Ensslen, F: Zwischenbericht aus dem Arbeitskreis Kantenfestigkeit im Fachverband Konstruktiver Glasbau e.v. (FKG). Glasbau 2013. Vol. 1. 2013. [2] DIN 1249-11: Flachglas im Bauwesen - Glaskanten, Begriffe, Kantenformen und Ausführung. Sep-1989. [3] preen thstr: Glass in building - Thermal Stress Calculation Method. Jul-2007. [4] DIN 18008-1: Glas im Bauwesen Bemessungs- und Konstruktionsregeln Teil 1: Begriffe und allgemeine Grundlagen. Dec-2010. [5] Lindqvist, M.; Vandebroek, M.; Louter, C.; Belis, J.: Influence of edge flaws on failure strength of glass. In Glass Perfomance Days. 2011. [6] Vandebroek, M.; Lindqvist, M.; Belis, J.; Louter, C.: Edge strength of cut and polished glass beams. In Glas Performance Days. 2011. [7] Sglavo, V. M.; Müller, C.; Righetti, F.: Influence of edge finishing on the resistance to thermal stresses of float glass. In Glass Perfomance Days. 2007. [8] DIN EN 1288-3: Bestimmung der Biegefestigkeit von Glas - Teil 3: Prüfung von Proben bei zweiseitiger Auflagerung (Vierschneiden-Verfahren). Sep-2000. [9] Mellmann, G.; Maultzsch, M.: BAM Forschungsbericht 161: Untersuchung zur Ermittlung der Biegezugfestigkeit von Floatglas für bauliche Anlagen, 1989. 10