O/H Sign failures at baseplate 8

[Dinosaur]

If I had a big financial grant I would study this and get a PhD or something, but since I don't I'd like to know if anyone has any thoughts on an intermittent failure we are having where steel posts are welded to baseplates and then failing in a brittle manner.

One of these situations exists where you have an Overhead Sign Structure supported by steel posts on each end. The posts are selected based on the total moment and shear at the base to determine the diameter and the wall thickness of the steel post. A pattern of anchor rods is selected to resist the same loads. The posts are welded to the base plates and the base plates are secured to the anchor rods with a nut top and bottom to provide a leveling device during installation.

The trick seems to be in joining the post to the base plate. To reduce the total volume of weld metal, many fabricators want to cut a circular hole in the base plate to fit the column into. A fillet weld is then made all around on the top side, and on the underside of the column to the inside of the circular hole in the base plate. I believe this method of connection has been forbidden in the latest Guide Specification. I have seen a fair number of cracks develop in this location in the HAZ of the circular column. Some have completely failed and appear to be a brittle mode.

I believe additional stresses are being introduced into the column due to the heat introduced into the column at the time of welding. The heat of welding causes the steel column to expand. The weld cools and fixes the column in a position above ambient temperature. After the assembly is completed, it is left with a high tensile residual stress acting circumfrentially at the weld, but because of the weld fixing the tube the stress is permanent. When the structure is placed in service, the design stresses act normal to the direction of this pre-existing stress condition. I believe the presence of these two substantial tensile stresses causes the column to fail in fatigue before it would be predicted if the circumferential stress were not present.

If anyone understands this rambling, let me know if you have thought about this problem before and what you think the cause may be.

 

[dimjim]

Is the hole baseplate chamfered before welding"
Is there a j groove on the pipe. From what I
have read, you simply have a clearance hole
in the base plate with 1/32 clearance. Bearings
are sometimes welded to a car body or car bed
and the circular diameter has a split in it
and it may be that it was for the reason that
you are implying. Can you split the pipe for a
short distance before welding? Are either parts
heated before welding? Can you weld the base
plates to the columns before assembly?
You might want to post this to the welding forum.
Are there specific specs for this type of welding
that you are using?

 

[Dinosaur]

Dimjim,

You are very enthusiastic about designing these supports, but that is not my question. These supports are designed by the contractor's engineer and reviewed by our structural department. I am not going to wade into the design. I left the structural design department because if I stayed, I would be doing these reviews and never get another promotion. Now I work in another department and I am looking at the specs to see if a better spec can be written to improve the life of these structures.

The poles are not perfectly round. The holes are not perfectly round. I would say that the combination of "out of roundness" results in a gap typically about one quarter inch, but sometimes more. As you can see from my earlier post, this results in a strain that is over 4x the strain necessary to yeild the pole. Is this a problem from the point of view of a restrained weld condition? If so, how can we spec the connection to eliminate the problem?

Currently, I can't advance this with the structural department because they don't see this as a problem. I don't think they understand the restrained weld problem and how temperature causes this strain.

 

[miecz]

According to Article 5.15.3 of the AASHTO Spec for Structural Supports of Highway Signs, Luminaires and Traffic Signals, the alternate connection is full-penetration groove weld. Using a solid base plate, a small hole in the base plate can be made for wiring/drainage.

What is the relative thickness between the pole and the base plate? I seem to remember that welding thin members to thick ones is results in greater residual stresses. Something to do with the thicker member acting as a heat sink, and the thin one getting extra hot.

Extensive experimental/finite element studies of tube to base plate connections have been made at Lehigh. These studies were mainly concerned with bolt stresses induced by the flexibility of the base plate. But flexure of the base plate would impact the weld as well. Cutting out the hole for the entire post would sgnificantly alter the flexibility of the base plate.

 

[Rentapen]

Most base plates and columns/poles are full penetration (small hole for wires and water) or slip fit (stronger-double fillet) but both are usualy rose bud heated to make a better weld. But to change the material to a brittle condition I would suspect over spec'ng the material/rod or the fab shop had some high strenght stuff laying around. Or bad location designing ( one size don't always fit all).
just wandering.
rent

 

[vmirat]

Dinosaur,

So far, I haven't seen an overwhelming response indicating that this is a universal problem, which makes me think you may be having a local issue with fabrication. As far as specs go, I suggest contacting AASHTO, AISC, and AWS to see if they can shed some light on this issue. Maybe this is a bigger problem than we know.

Something else occurred to me. If they're welding the pipe at the same location above and below the base plate, does this increase the size of the HAZ for both the plate and pipe? Maybe they could weld a ring sleave to the base plate first and then insert the pipe support into the ring sleave and weld at the top of the sleave. That way, you get some separation between the two welds. Just brain storming here.

 

[dimjim]

We were having trouble with some weld materials
cracking due to too much sulphur in the parent
materials. Controlled the sulphur in our specs
and the problems went away.

 

[Dinosaur]

One reason I am posting the question here is because I think I need structural engineers to contemplate the problem with me. Welding guys I ask don't understand the relationship between the heat and the strain, but these guys are not engineers. The welding guys are looking at the C.E. for the steel and the weld rod, the need for preheat to eliminate HAZ concerns and this sort of stuff. All of this is good for improving the quality but does not zero in on my concern.

Structural engineers understand poisons ratio, and the significance of the circumferential strain being 6x the yeild strain, and the potential stress caused by double curvature bending through the wall thickness as it transitions from an artificially large radius back to the "at rest" radius, and that if all these problems could be cause by 200 deg F then it is very likely to be a problem.

I have been brain storming this off and on for about two years. Every solution I think of sounds too expensive. Checking into sulfer content had not occured to me except that it is addressed somewhat in the C.E. formula.

There is an engineer that works for a fabricator that I believe has all the right skills. I plan to chat him up next time I am in that shop.

Anyway, I appreciate the help.

 

[MWPC]

Forensic building engineers have found lots of brittle failures after major earthquakes. Many are caused by the steel being restrained in the orthogonal directions. I.E. it could not yield by necking down. I did not see a reply to the above question about the relative thicknesses of the pole and the plate but if the failure is right at the top of the top fillet weld then I believe restraint may be a contributing factor along with the residual welding stress and possibly some bending due to the anchor bolts not being in line wth the pipe wall. Exactly where is the failure? On those that have only cracked do the cracks show up near the bolt locations? Ken

 

[Dinosaur]

The failures are usually in the weld or the piece of the pole extending from the weld, maybe up to 1 inch away from the weld.

The pole has a comparatively thin wall thickness, probably up to 3/8 to 1/2 inch, but may get thinner than 1/4 inch.

The base plates are usually from one to two inches thick.

The anchor rods are a minimum of four-one inch diameter anchor rods of A36 steel. We have had anchor rods fractured in the top third and we have had anchor rods work loose in the concrete.

I don't recall the specific grade of steels. I believe it varies from one manufacturer to another. It depends on where they go to get their supply of steel. Our requirements are for a minimum yeild of 36 ksi, an 18% elongation in the test specimen, a C.E. below 0.45, and soon a Charpy V-notch test.

 

[MWPC]

Dino,

Without reviewing any of the calc's, my best guess is that the designer is ignoring the local bending due to the anchor rods not aligning with the shell of the pipe. In building design the base plate is continuous and you check bending in the plate due to the distance from the flange of the column to the center of the bolt. In your case the plate is not continuous therefore the bending has to be taken by the wall of the coulmn or the anchor rods. Without grout there will be stress reversals. You seem to have had failures at both locations.

Can you convince your structural engineers to check for this in the shop drawing review? Can you specify that the loads from the pipe be transfered with stiffeners without counting on the circumferential weld? It may cost a little more but if they are all bidding the same spec no one should complain. Good luck.

 

[Dinosaur]

MWPC,

The forces and moments causing load in the anchor rods are a result of the forces and moments used to design the poles. It seems the forces and moments in the pole have to have been checked.

In a building column, you have an axial force and moment in the column, and the base plate is checked against the effects of the eccentric anchor rod to the column flange tip usually.

Unless you are alluding to a likely unequal load distribution in the pole due to the flexibility of the baseplate transfering four point loads into the pole, which I believe would have to be checked using a fairly rudimentary FE model, I don't get the disconnect between the design of the pole and the design of the anchor rods being non-conforming.

Stiffeners have proven to be hard points that increase the likelyhood of introducing fatigue problems.

Maybe I'll have to ask our Structural department to perform a few representative FE model analases. Unfortunately, due to personalities and budget issues, that will be a long time obtaining.

It is my suspiscion that the failures in the anchor rods are largely due to poor construction practices.

 

[MWPC]

Dino,

All I am saying is that there is an eccentricity between the wall of the column and the centerline of the bolt. If the base plate is not grouted to provide some fixity, this local bending must be taken in the bolt and/or the the wall of the column. I am not an expert at sign base design. Maybe designing for this local bending is SOP and is in the calcs. I think it would be worth a quick check. I have seen much worse overlooked by designers.

 

[Dinosaur]

When I was designing O/H supports, I accounted for the bending in the anchor rods due to the ungrouted condition since I knew the agency would not permit grouting. I don't think many were as rigorous. I have also seen much more significant mechanisms overlooked by designers.

I am currently fixated on the cracks at the pole base.

I wish I had the resources to prepare a quick FE model to look at some of these other effects.

 

[VTPE]

Perhaps you are seeing a local effect from an abrupt change in stiffness of the resisting material. In theory, when you weld (2) pices of steel together, they become one. Following this theory, the cross section of the element at the base plate would be orders of magnitude stiffer than the pipe directly above the base plate. The abrupt change in stiffness equates to an abrupt change in stress gradient. Tapered stiffeners tend to smooth out the stress gradient to the base plate. The doctorial thesis would be to examine the finite element analysis on lengths of stiffener required to transition the stresses within an acceptable range. Good luck.

 

[miecz]

In addition to your concerns about residual shrinkage stress, the double fillet weld would seem to load the upper weld more than the lower. Under a static load condition, it would be safe to assume that, at ultimate, the two welds share the load. But under a stress reversal, I expect the upper weld is going through larger stress cycles.

If fatigue is the concern, than I would think the groove weld allowed by AASHTO Article 5.15.3 would be a better option.

 

[dimjim]

happened across this on another post.

Fatigue and Fracture Control

As a crystalline material, steel is very susceptible to brittle fracture due to fatigue. There are other mechanisms that lead to brittle fracture as well. The SCM discussion on pg 2-33 is a must read section for engineers designing in steel.

Wind and Seismic Design

Both wind and seismic events have very high strain rates which can lead to brittle behavior. In addition, current practices in seismic design typically do not design structures to resist forces elastically. There is an inherent dependence on material ductility in the design philosophies. Special detailing is required to ensure ductile behavior of steel structures in these events. There has been significant advancements in this area in recent years. These detailing requirements, again, are topics for a more advanced course, however, you should take the time to read the SCM discussion on pg 2-35

 

[Dinosaur]

jmiec,

Yes, I agree the upper weld is receiving more load but only a working stress methodology will reveal it. Unfortunately, most engineers are only learning ultimate strength/plastic design concepts so they don't perceive this. My thinking on this is that if the upper weld has to yeild to permit the lower weld to share the load, then under fatigue considerations the upper weld is undergoing too much strain.

The problem with a groove weld is how do you do it? You can't groove weld it from both sides, so you have to do it from one side only. This requires the backing bar to remain in place which is a very poor fatigue performer. And then when it is over, how do you inspect the weld to insure its soundness? I doubt you can radiograph it and ultrasound will probably pick up the installed crack at the backup bar.

 

[miecz]

According to Table 11-2 of AASHTO's Luminaire Spec, the groove weld (Ex. 5) with the backer gets a fatigue category E`, same (to my surprise) as the fillet welded connection (Ex. 7.)

The groove welded connection can be upgraded to category E by attaching the backer ring to the plate with a full penetration groove weld.

Even though the stress category is similar to the fillet welded connection, the stress range in the groove weld will be much less since there's a lot more weld there, especially when compared to the upper weld of the double fillet detail.

 

[Dinosaur]

Back to the original question, do the stresses induced on the column due to the restrained weld cause any concern among the rest of you fellows?

 

[BigInch]

I don't think that should be a concern, if post weld treatment is used where the material thickness would so require.

It sounds like a connection detail problem to me. The only weld I would use in this situation is a full penetration bevel weld from column to plate without cutting a hole in the plate. If you don't have time to put that on a drawing, outsource the design.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com