The Glue Strength Test: Results

We conducted the glue strength test today using a new set of test articles designed to fail well under the maximum load capable of being applied by a tensile strength test machine here at the University of Minnesota. My cohorts in this project were undergraduate engineering student Aaron Bardon, and Professor Jeffrey Schott of the Chemical Engineering and Materials Science Department at the University. I said in my previous entry that my assumptions and calculations were simplified, and that part failure would probably be based on a number of modes of failure rather than on one simple cause. The results were surprising to us to say the least.

The results are summarized here. I should state up front that no particular glue or brand is advocated or disparaged here. Different varieties of glue bring their own unique characteristics to a project, and this test is simply one of strength comparison when applied to a set of identical parts constructed of quartersawn red oak with mortise and tenon joinery.

The three varieties of cyanoacrylate glues will be discussed first since this project was designed to test a new brand of these glues specifically formulated for commercial woodworking applications against the traditional water-based polyvinyl acetate wood glues. These are marked four through six in the photo here, and their test results are summarized below.

4. Cyanoacrylate glue with short open or working time. The glue failed at around 2000N or about 450 lbs. of force in tension.

5.  Cyanoacrylate glue with medium open or working time. The glue failed at around 2400N or about 540 lbs. of force in tension.

6. Cyanoacrylate glue with long open or working time. The glue failed at around 1800N or about 405 lbs. of force in tension.

The important thing observed here is that all three varieties of cyanoacrylate glue failed rather cleanly when a force similar in magnitude was applied, and failed before the wood did. I predicted that the wood would fail in shear first, not the glue. This was not the case though.

Two common water-based polyvinyl acetate glue formulations are marked one and two respectively in the accompanying photo here. Test results are summarized below.

1. PVA Type-I woodworking glue. This part did not fail even at the maximum test force of 5000N or 1124 lbs. applied in tension.

2. PVA Type-III woodworking glue. The part containing the mortise itself failed parallel to the grain of the wood approximately at the mortise depth. The section containing the mating tenon remained intact. The glue joint did not fail. The wood failed either in tension or shear perpendicular to an applied force of 3900N or 877 lbs. A stress concentration at the mortise probably contributed to failure here.

I included two special case parts for comparison, and their test results are described below.

3. Polyurethane glue. The glue failed at around 2000N or about 450 lbs. not too unlike the three cyanoacrylate glues. I included this glue in the test because it is specifically promoted for its high strength. In this test though, not so much.

We concluded the test with the part marked #20, and any woodworker knows what's coming. The part was joined using a #20 biscuit glued with PVA Type-I woodworking glue. What most woodworkers would not expect is that the part shown in the photo did not fail when tested to 5000N or 1124 lbs. I now have a much healthier respect for biscuit joinery, and would consider using a biscuit or two for strength rather than simply for alignment as is usually the case in my shop.

The cyanoacrylate glues however, do not get that same respect. They might have been useful to me in the construction of an organ case, but case frame members have to carry heavy loads, and these glues appear to be significantly inferior to the polyvinyl acetate glues with respect to strength. Although all joint tolerances were kept close, the water-based polyvinyl acetate glues may in the end have superior bonding strength because they are capable of swelling the joint. Because they do not cure within minutes, the polyvinyl acetates may also penetrate the wood pores deeper resulting in a better and stronger bond.

We debated these ideas as we concluded the test, and to be honest, this is where the fun in all this is. You make a guess about something and test it. Sometimes you get it right, but sometimes the test can lead you in a completely new direction. I learned a lot today.

Much thanks goes to Professor Jeffrey Schott and Aaron Bardon at the University of Minnesota for graciously agreeing to be part of this test project.

Bibliography.

Higdon, A., Ohlsen, E., Stiles, W., Weese, J., & Riley, W. (1976). Mechanics of Materials. New York: John Wiley & Sons.

Hoadley, R. Bruce. (2000). Understanding Wood. Newtown: The Taunton Press.

The Glue Strength Test: Engineering Test Articles

I introduced the glue strength test project in my previous entry. Working with an engineering professor and undergraduate student at the University of Minnesota, the project is designed to test a new woodworking glue against traditional glues by tensile strength testing a series of identical mortise and tenon joints to failure. The problem with the initial test was that the part was built too strong for the test machine which could not exert enough force to cause it to fail.

I decided then to apply engineering to the design of a new part that would hopefully experience failure, and therefore provide a meaningful test of one glue against the other. I had to make assumptions about where failure would occur to model the modes of failure most likely to occur, and decided that the glue could fail where the mortise and tenon were joined, the wood itself could fail at the tenon due to the force in tension being applied by the test machine, or the wood could fail at the tenon due to the shear force applied by the test machine.

I began by looking up the equations and published strength data for both the glue and wood used in the test samples pertaining to the three modes of failure I identified as possible. What I found was that there was little published data on the strength property of wood in tension parallel to the grain as this number was always much greater than other strength property figures for other modes of failure of wood in general. The more likely modes of failure then were going to either be failure of the glue, or failure of the tenon in shear.

I set up the document shown here in the iPad app PocketCAS pro to calculate the force at which a joint would fail due to both glue failure, and failure of the wood due to shear force. My goal was to design a joint that would comfortably fail within the test machine force limit of 5000N or 1124 lbs. There is an axiom among woodworkers that glue is stronger than wood, and the equations prove this true. The wood will fail in shear before the glue will fail according to the math, and I designed the tenon dimensions to comfortably fail well under the maximum force the test machine is capable of exerting on the part assembly.

An interesting find obtained by the mathematical model is that such a small tenon can withstand a force up to 667.5 lbs. due to the shear strength of red oak parallel to the grain. I expect this number could be different by some because there may not be a single and simple mode of failure, but rather failure may be multi-modal. In other words, the glue that joins mortise to tenon may play a role in how the joint fails.

I sort of hope this is the case because that would mean that the strength properties of different glue types factor in, and are somewhat represented in the test result data obtained. If so, their effect on joint strength will factor into my work. If not, there is no strength restriction placed on the type of glue I can use in any one construction.

A last word about my assumptions and corresponding calculations. They are simple, and do not represent anything near the complex mathematical models that may more accurately describe the behavior of the glued mortise and tenon joint under tension. I am at this point looking for a ball-park figure so that we can get one of the assemblies to fail during test. An analysis of test results will later provide a more accurate direction to take with regard to the interactions between force, glue, and wood in this type of joint.

The Glue Strength Test: Introduction

I recently came across an article in a trade journal about a company that had begun producing a new glue for commercial woodworking applications that promised quick dry time thereby negating the lengthy times normally associated with clamping glued parts. The new glue is a cyanoacrylate glue, and is far different from the traditional polyvinyl acetate woodworking glue I normally use. Consumer formulations of cyanoacrylate glue are commonly referred to as super glue, or simply as CA glue.

The quick dry time got me interested though because there are applications where this could be useful if the glue proved to be strong enough. I contacted the company making the new glue for more information, and they provided me with a sample pack of the three varieties they offer differing only in the open or working time of the glue, short, medium, and long.

I also happened to recently meet engineering student Aaron Barden who specializes in materials science at the University of Minnesota. Materials Science at the University is part of the Department of Chemical Engineering. Aaron put me in touch with Professor Jeffrey Schott who agreed to be part of a tensile strength test using the new glue against the traditional glue in parts joined from each.

I decided to make six t-shaped parts using mortise and tenon joinery for the test such that two were glued with commonly used polyvinyl acetate glues, one with polyurethane glue, and the remainder with the three varieties of the new cyanoacrylate glue. I chose a tenon size that seemed typical for a normal frame design. With six test articles built, we set about to test the strength of the glue joints.

Testing was done on the tensile test machine shown above. A part is clamped in the machine which applies a force that increases steadily over time, and because most materials stretch some when a force in tension is applied, the machine plots that force against the change in length of the part until ultimate failure occurs. We didn't expect the wood to stretch much over the course of the test, but we did expect the joint to fail before the maximum force limit of the machine was reached. The joint astonishingly did not fail under the maximum force applied to it of 5000 Newtons, or about 1124 pounds. We agreed to run the test again with new parts designed with smaller joints that would actually fail.

So walking back after that initial test conducted during the lunch hour, I concluded that I needed to understand the shear strengths of both wood and glue involved, and make an attempt to apply some real engineering to the problem. There are some surprising numerical results for the size of a joint that should fail under the load parameters capable of being produced by the test machine, and I will publish those in my next entry.

The next step is to produce new parts based on the numerical results obtained from published data on the shear strength of both glue and wood, and then test those parts to catastrophic failure. In doing so I can get an idea of the relative strength of the new glue, and be comfortable using it as a substitute method where appropriate.