Structural Repairs and Maintenance of Heritage Architecture XII285Towards a non-destructive methodology todistinguish wrought iron from mild steelin 19th century structuresI. Wouters1, I. De Graeve2, D. Van de Velde1,M. de Bouw3,4 & Q. Collette11Vrije Universiteit Brussel,Department of Architectural Engineering, Belgium2Vrije Universiteit Brussel,Research Group Electrochemical and Surface Engineering, Belgium3Artesis University College Antwerp, Department of ArchitecturalSciences – Conservation of Monuments and Sites, Belgium4Belgian Building Research Institute, Lab of Renovation, BelgiumAbstractDuring the 19th century various new construction materials became available in ashort time. This paper deals with the quest for a methodology to differentiatewrought iron from mild steel by using a combination of several onsite nondestructive testing instruments. A mobile Vickers hardness tester and an opticalmicroscope were used to determine the hardness and analyze the microstructureafter onsite polishing and nital etching of historic wrought iron and mild steelstructures. The historic specimens were also tested in a destructive way (tensiletest) to obtain values for the mechanical properties and relate these values to theNDT results. Although the hardness measurements showed very large scatter, atrend in the measurements could be defined: large variations in local hardnessmeasurements are a clear indication of wrought iron, which can be explained bythe inhomogeneous micro-structure. Low variation in hardness is typical for mildsteel. In the latter case a conversion from hardness to tensile strength is possible.The obtained dataset, coming from different Belgian structures dating from 1895to 1905, is compared to datasets originating from the UK and US in order toposition the Belgian historical iron alloys within a larger international context.Keywords: hardness test, wrought iron, mild steel, metallography, NDT.WIT Transactions on The Built Environment, Vol 118, 2011 WIT, ISSN 1743-3509 (on-line)doi:10.2495/STR110241

286 Structural Repairs and Maintenance of Heritage Architecture XII1 IntroductionDuring the second half of the 19th century various new construction materialsbecame available in a short time. Engineers first applied cast iron to createslender structures, mostly in compression. Around 1860 they turned to wroughtiron and from 1880 onwards to mild steel.Where cast iron can easily be distinguished with the naked eye due to its formand surface structure, the difference between sound wrought iron and mild steelis not visible as the same techniques were used to roll sections into their finalform. However, when focusing on the production process, the microstructure andthe mechanical properties, wrought iron and mild steel are clearly two differentproducts and one should take this difference into account when renovatingstructures.Previous research [1] has indicated that using the terminology ‘wrought iron’and ‘mild steel’ is confusing. In this paper we will talk about weld iron/steelwhen we refer to the inhomogeneous fibrous product that has been produced bythe puddling, the reverberatory, the dansk or the rotary furnace. We will talkabout ingot iron/steel when we refer to the homogeneous product that comes outof the Bessemer/Siemens Martin or Thomas converter.1.1 Weld Iron/steelWeld iron/steel has a layered structure. It is composed of thin layers of almostpure iron with thin threads of slag in between. The large slag elements,sometimes visible with the naked eye, are squeezed into tubes due to the rollingprocess. The orientation of the slag in weld iron/steel causes differentcharacteristics in the transverse and longitudinal direction. The strength in thelongitudinal direction of a bar is on average 7 to 10 percent higher than in thetransverse [2, 3]. Analysis of historical test data and modern tests on historicalsamples show that the variation of mechanical properties (tensile strength,ductility, toughness, elongation) is quite high between structural sections andeven within a single section [4, 5]. Weld iron/steel can be weak or strong (tensilestrength 140 – 530 MPa), brittle or ductile (elongation between 1 and 36percent).1.2 Ingot iron/steelIngot iron/steel has been heated to a liquid form during the manufacturingprocess. It has undergone fusion leading to a homogeneous structure. Ingotiron/steel has a more consistent microstructure then weld iron/steel. Theimpurities or inclusions in the steel are much smaller and more distributed.2 Test set-upIn general, when renovating a metal structure, a sample is taken from thestructure and tested in a destructive way. As it is not always possible or desirableWIT Transactions on The Built Environment, Vol 118, 2011 WIT, ISSN 1743-3509 (on-line)

Structural Repairs and Maintenance of Heritage Architecture XII287to remove samples from a structure, we investigate what kind of informationnon-destructive tests can deliver. Nowadays onsite chemical analysis, hardnesstests, and metallography are possible. This paper examines the use of a nondestructive hardness tester as well as metallographic analysis.As the results obtained by non-destructive testing have to be evaluated withthe real material characteristics, the samples were also tested in a destructiveway. Five samples from structural elements (I- and U sections, plates) wererecuperated from renovated Belgian buildings dating from 1895-1905.2.1 Destructive testingTensile testing coupons were machined from the structural elements, accordingto EN10 002-1, to determine the strength and ductility.Table 1:Mean values of the Young’s modulus, yield tress, ultimate tensilestrength, strain, area reduction and Poisson’s ratio for five historicsamples determined by tensile 30538037120172926-0,29-0,29Remark: ‘2/1895/U’ refers to sample 2 cut out a U-section from a 1895 building2.2 MetallographyTo study the microstructure’s grain size and inclusions, the samples weremechanically polished and nital etched in lab conditions, as preparation foroptical microscopy analysis. To compare this mechanical lab-etchingmetallographic procedure to the onsite metallographic sample preparation, onsiteconditions were also imitated in the laboratory. The surfaces of the metalsamples were polished manually by gradually diminishing the grade of wet(silicon carbide) sandpaper from P#500, P#800, P#1200, P#2400 to P#4000 (seeFigure 1). Subsequently a small area of the polished sample is chemically etchedby dribbling nital on the cleaned surface for 5 minutes.The same optical microscope was used to study the lab as well as the onsitesamples, but magnifications were limited to values feasible for onsite analysis.Magnifications of x50, x100 and x200 were applied when analysing the etchedsamples. The pictures from the lab etched samples are sharp. The homogeneousstructure of sample 3 refers clearly to the ingot-procedure, whereas the largeWIT Transactions on The Built Environment, Vol 118, 2011 WIT, ISSN 1743-3509 (on-line)

288 Structural Repairs and Maintenance of Heritage Architecture XIIFigure 1:Manual polishing of the metal sample (left). Manual nital etchingof a defined area on the metal sample (right).inclusions, typically elongated in one direction, and the layered structure ofsample 4 refer to the weld-procedure.Although the pictures of the onsite etched samples are less clear due to highersurface roughness of the manual polishing, this trend is visible on a scale of 200xas well as on a scale of 50x, which is possible to apply onsite.As a consequence, onsite polishing, etching and metallography can be used todetermine whether steel structural elements were produced according to theweld- or ingot-procedure. The metallography does not give any informationabout the strength or ductility of the sample.2.3 Portable hardness testerHardness tests are extensively used in quality control. The measurements are fastand easy to perform, which make them attractive to use during a renovationprocess. For modern steel there is a reasonably accurate correlation betweenhardness and tensile strength and conversion tables available, which are based onnumerous tests [6]. The validity of this correlation for historic weld and ingotiron/steel structures remains to be determined.Different instruments can be used to measure the hardness. As themicrostructure of the historical metal is heterogeneous, the diameter of theindentation will influence the measurement. The larger the size of the indenter,the more average the measured hardness will be, as it is measured on a surfacearea containing various microstructural inclusions. Conversely, when smallerindentations are used, like with a Vickers test, more local hardness information isobtained, which will reveal influences of local surface differences inmicrostructure and enables to reveal a hardness gradient.The hardness of the five polished samples was determined using three statichardness testers:-Rockwell hardness tester, scale B, 500N load, indentation 1/16 inchsteel ball,Vickers hardness tester ‘Struers Duramin’, load of 20N during 10s,Portable MIC 10 Vickers hardness tester, with 205-A indenter (50N).On every historic sample 10 hardness measurements were taken for eachhardness test. The loads applied on the metal samples give indentations of aboutWIT Transactions on The Built Environment, Vol 118, 2011 WIT, ISSN 1743-3509 (on-line)

Structural Repairs and Maintenance of Heritage Architecture XIILabo etching289Onsite etchingSample 3/1905/I(ingot-procedure)X50X200Sample 4/1903/(weld-procedure)X50X200Figure 2:Comparison between the pictures of the metallography(magnification x50 and x200) from the mechanically (left) and‘onsite’ (right) polished samples for the ingot (up) and weld (down)procedure.150μm with the Vickers tester, 270μm with the portable Vickers tester and950μm with the Rockwell tester. Figure 3 shows the proportion of theindentation compared to the slag (50μm).Table 2, which gives the mean values and the standard deviation of thehardness measurements, illustrates that the standard deviation is larger for thesamples fabricated according to the weld-procedure. Hence, we can state thatlarge deviations of the hardness measurements indicate a heterogeneous structureand point to the weld-procedure, whereas small variations point to the ingotprocedure.WIT Transactions on The Built Environment, Vol 118, 2011 WIT, ISSN 1743-3509 (on-line)

290 Structural Repairs and Maintenance of Heritage Architecture XIIFigure 3:Schematic of the influence of the indentation area of the hardnesstesters (left) on the analysed surface of historic ingot iron/steel(middle) and weld iron/steel (right) with different microstructures.Table 2:Hardness of the five historic metal samples tested with (portable)Vickers and Rockwell hardness testers.SamplesVickersHV 20NPortable VickersHV 50NRockwell BHRB edure4/1903/5/1903/-17412420112101205344736376The prediction of the ultimate tensile strength for the weld procedure showslarger variations. Other researchers were confronted with the same conclusions[4, 5, 8]. Bowman and Piskorowski [5] collected historical data sets of tensileload tests on wrought iron. Figure 5 plots the ultimate tensile strength of morethan 1500 wrought iron bars, plates and angle iron tested by the English scientistKirkaldy and the American Beardslee together with the Belgian specimens data.To convert the hardness values into ultimate tensile strength, we usedUconeer [7], which is a program based on an extensive set of experiments onmodern carbon steel and steel alloys, as conversion tables for historic steel donot exist (yet).Table 3 indicates that the ultimate tensile strength could be predicted from thehardness values within a margin of 11% for ingot samples. This might not beaccurate enough for a structural calculation, but will enable engineers todetermine whether the steel qualities of two structural elements are comparable.Figures 4 and 5 show high variation of the tensile strength and percentelongation for historic wrought iron. The Belgian specimen show higher tensilestrength but lower ductility, which indicates once again the importance todetermine the mechanical properties when dealing with weld iron/steel.WIT Transactions on The Built Environment, Vol 118, 2011 WIT, ISSN 1743-3509 (on-line)

291Structural Repairs and Maintenance of Heritage Architecture XIITable 3:Conversion of hardness to ultimate tensile strength (UTS) andcorrelation with tensile test.SamplesUTS converted from hardness testPortableVickersVickersRockwell BHV 20NHV 50NHRB 9680396446431123792550Ultimate tensile strenght [MPa]500450400350300250200150100Figure 4:TestsamplesBegian bar 1903Kirkaldy (bar)Kirkaldy (angle)Beardslee (bar)Kirkaldy (plate)Ultimate tensile strength of the Belgian coupons against the historicdataset on wrought iron by Kirkaldy and Beardslee [5].WIT Transactions on The Built Environment, Vol 118, 2011 WIT, ISSN 1743-3509 (on-line)

292 Structural Repairs and Maintenance of Heritage Architecture XII40Percent elongation [%]35302520151050TestsamplesBelgian bar 1903Figure 5:Kirkaldy (bar)Beardslee (bar)Percent elongation of the Belgian coupons against the historicdataset on wroug