Destructive bend test, three beam types

Engineering test poster

An engineering poster is read the way a test report is read: rig, standard and tolerances first, then what broke, at what load and how. This A0 landscape example reports a fictional four point bend series on twelve timber beams, with the rig as a five entry protocol, one load against deflection chart, every specimen load printed in the results text and a key finding callout.

Create a poster with OneCraftA0 Landscape, printed at 1189 by 841 millimetres

The whole board

The poster at full size, exactly as it prints. Every number, citation and caption on it was written for this example, so the layout is being judged on real content.

Screwed steel flitch plates raise the failure load of 240 mm timber beams by 48 percent

O. Baptiste ¹, H. Sato ¹, G. Ferreira ²

1 Structures Laboratory, Corriedale Institute of Technology · 2 Corriedale Timber Products

ABSTRACT

Twelve 240 by 90 mm laminated timber beams were loaded to failure in four point bending over 3.6 m: four timber only, four with a glued 6 mm steel flitch plate, four with the same plate screwed at 150 mm pitch. Mean failure loads: 28.4, 36.9 and 42.1 kN. Screwed plates carried 48 percent more than timber alone, and two of four glued beams failed early by bond slip, the finding that decides how these beams are built.

INTRODUCTION

A flitch beam sandwiches a steel plate inside a timber section so a renovation can carry new load without a deeper beam, and the steel works only while it stays composite with the timber. How the plate is fixed is therefore the whole question. Glue gives the stiffest connection on paper and the least forgiving one in practice: a glue line cannot be inspected after assembly, cures badly in cold or dusty site conditions, and when it lets go it lets go at once. Screws are slower to install and leave visible heads, but every one of them can be checked, and a screwed connection sheds load gradually rather than suddenly. Published capacity tables for flitch beams in this size range assume full composite action and say nothing about which fixing method delivers it at failure.

METHODS

Specimens. 240 x 90 mm laminated timber, 3.6 m span, four per type: timber only, glued flitch, screwed flitch.

Rig. Four point bend with load points at the third points, 250 kN actuator, two LVDTs at midspan for deflection.

Rate. Displacement control at 2 mm per minute, loaded monotonically to failure; load and deflection logged at 10 Hz.

Standard. Test arrangement and specimen conditioning followed the national timber structures bending test standard cited below.

Tolerances. Plate thickness 6.0 mm plus or minus 0.1 mm; screw pitch 150 mm plus or minus 2 mm; span set to plus or minus 5 mm.

CONTRIBUTIONS

The first side by side failure mode comparison of glued and screwed 6 mm flitch plates at a 3.6 m span, with every specimen load and failure mode reported so each mean and the recommendation can be rebuilt.

Figure 1. Load against midspan deflection, one representative beam per type. Glued and screwed share a slope; screwed fails highest at 42 kN.

Source: four point bend series, twelve beams, 2026.

RESULTS

Mean failure loads were 28.4 kN for timber only (27.8, 29.1, 28.0, 28.7), 36.9 kN for glued flitch (35.6, 38.0, 37.2, 36.8) and 42.1 kN for screwed flitch (41.2, 43.5, 42.0, 41.7). Screwed plates carried 48 percent more than timber alone (42.1 / 28.4 = 1.48) and glued plates 30 percent more. Two of the four glued beams failed by bond slip below the glued mean, G1 at 35.6 and G4 at 36.8 kN. Initial stiffness was within 6 percent between glued and screwed beams, so the fixing showed itself only at failure. Nine of twelve failures started on the tension face, two were those bond slips, and the twelfth was screw yield at 43.5 kN, the highest load recorded in the series.

DISCUSSION

The number that matters most is not a mean. Two of four glued beams failing by bond slip means the glued connection, not the steel or the timber, set their capacity, and a site assembled glue line has every disadvantage the laboratory one did not. The screwed group's range of 2.3 kN was close to the glued 2.4, timber alone was tightest at 1.3, and the screwed group's one non tension failure was ductile screw yield at the top of the series.

Key finding

Screwed steel flitch plates carried 48 per cent more load than timber alone, 42.1 kN against 28.4 kN.

Screwed plates: 42.1 kN mean failure load

Glued plates: 2 of 4 failed by bond slip

Timber only: 28.4 kN mean

Recommend screwed plates for site assembly

REFERENCES

1.Baptiste O, Sato H. Four point bending of composite timber steel sections: rig design and measurement. Structures Testing Journal 2024;19(2):101-116.

2.Ferreira G, Baptiste O. Flitch beam behaviour under sustained and short term load. Timber Engineering Notes 2025;8(1):40-56.

3.Sato H, Marchetti V. Reporting destructive test series: a checklist for structures laboratories. Laboratory Practice in Engineering 2023;5(3):177-190.

Block by block

What each block on the board is for, in the order a reader walks it.

Title band and authors: The title is the design recommendation with its number: screwed steel flitch plates raise the failure load of 240 mm timber beams by 48 percent. Three authors sit under it, and the timber supplier appears as the second affiliation, which is where a test series like this usually gets its specimens.
Abstract and introduction: The abstract carries the series in four sentences: twelve beams, three types, a 3.6 metre span, mean failure loads of 28.4, 36.9 and 42.1 kN, and two of four glued beams failing early by bond slip. The introduction frames the choice: glue gives the stiffest connection on paper and the least forgiving one on site, because a glue line cannot be inspected after assembly, while every screw can be checked.
Methods: specimens, rig, rate, standard, tolerances: Five protocol entries: specimens, the four point rig with load at the third points, a 250 kN actuator and two midspan LVDTs, displacement control at 2 mm per minute logged at 10 Hz, the national bending test standard followed, and tolerances, plate thickness 6.0 plus or minus 0.1 mm, screw pitch 150 plus or minus 2 mm and span within 5 mm.
Load against deflection (Figure 1): One representative curve per beam type rather than twelve overlapping lines. The caption states the reading: glued and screwed beams share a slope, and the screwed beam fails highest at 42 kN. Three lines keep the shape difference readable at a metre, and the shared slope makes the stiffness point without extra prose.
Results: every specimen in text: There is no table on this layout, so the results block prints all four values behind each mean: timber 27.8, 29.1, 28.0 and 28.7 kN; glued 35.6, 38.0, 37.2 and 36.8; screwed 41.2, 43.5, 42.0 and 41.7. The means recompute exactly, and 42.1 over 28.4 is the 1.48 behind the headline. The text also reports stiffness within 6 percent between glued and screwed beams.
Failure modes that add up: The mode sentences agree with the values. The two bond slips, G1 at 35.6 and G4 at 36.8 kN, are the two lowest glued loads, and both sit below the glued mean of 36.9. Nine tension failures, two bond slips and one screw yield make the twelve beams tested. The discussion gives each group's range, timber 1.3, glued 2.4 and screwed 2.3 kN, so no group is called more consistent than its numbers allow.
Discussion, key finding and conclusion points: The discussion argues that the mode matters more than the mean: a fixing that fails by bond slip in half its specimens sets the recommendation whatever the averages say. A key finding callout repeats the 48 percent result, and four conclusion points give the screwed mean, the glued bond slips, the timber mean and the recommendation for site assembly.
Contributions and references: The contributions block says what is new: the first side by side failure mode comparison of glued and screwed plates at this span, with every load and mode reported. The three references are invented for this fictional study because the layout carries a reference block, and the figure's source line names the test series.
How to adapt this board: Put every specimen value in the results text, grouped by type, with its failure mode beside it, so each mean can be rebuilt. Chart one representative curve per type, cite your test standard in methods, print tolerances, and count failure modes so they total the number of specimens tested.

What makes this board work

Every mean can be rebuilt

All twelve failure loads are printed in the results text, four per type, and each mean on the board is their average. A reader with a phone calculator can confirm 28.4, 36.9 and 42.1 kN and the 1.48 ratio standing at the board.

One curve per type

Figure 1 shows three load deflection lines, not twelve, so the shared slope of glued and screwed beams and the higher screwed failure load read from across an exhibition hall.

Tolerances printed

Plate thickness 6.0 plus or minus 0.1 mm and screw pitch 150 plus or minus 2 mm sit in the methods block. Tolerances are the first thing a test reviewer looks for and the first thing most posters omit.

Questions people ask

What goes on an engineering poster?

The rig, the standard followed, the tolerances, the result for every specimen, one chart, and a recommendation sized to the evidence. This board ends at a fixing recommendation rather than a capacity claim, because twelve beams from one timber batch can support a choice between two fixings but not a design table for every span.

Where do individual specimen results go without a table?

In the results text, grouped by type, one list of values per group. This layout has no table block, so the board prints all twelve loads as three lists of four. Put the failure mode next to each value too; a mode count given only as a summary sentence is where errors creep in. This board's modes total the specimens: nine tension, two bond slip, one screw yield.

Which chart suits load against deflection?

A line chart with one series per specimen type. The figure block accepts ten chart kinds, and line is the one that shows stiffness as slope and ductility as length of run. Twelve individual curves would be honest but unreadable, so one representative curve per type is the usual poster choice, with the caption saying which specimen it is.

Should I show a photo of the rig?

Not on this board. Its figure is a chart, and the photo figures on the four image based poster layouts have no upload control, so they cannot be filled. Describe the rig in a methods entry instead: four point bend, load at third points, 250 kN actuator, two LVDTs at midspan. Link rig photographs from a printed address.

Can I cite a standard on the board?

Yes, twice: name the test standard in the methods protocol where the procedure is described, and put the full citation in the reference list. State the loading rate and specimen conditioning as well, as this board does with 2 mm per minute displacement control, so a reader can judge conformity rather than take it on faith.

Is landscape right for engineering results?

Landscape suits a board with several columns of short blocks read left to right, like a test report: methods, figure, results and discussion. This example is on the A0 landscape canvas at 2340 by 1656. A portrait board suits a single dominant figure with the text stacked above and below it.

Build your own in about a minute

The button below opens the generator with this use case already described. Change the wording to match your own, generate, then edit anything you like.

Make one like this

Sources

Written and checked by the OneCraft team. Last checked 29 September 2026.