Return to index: [Subject] [Thread] [Date] [Author]

Re: QUERY: Fatigue

[Subject Prev][Subject Next][Thread Prev][Thread Next]
Bill:

I presume the cover is directly exposed to wheel load, i.e. there is no material
above the cover. If there is some material (i.e. the vault is buried) you may take
advantage of load dispersion in computation of stresses.

There are two types of fatigue failure: (a) material failure and (b) Weld failure.

Material fatigue failure is checked with the help of Goodman Diagram or similar.
Normally, as a designer, I would like to keep the extreme fibre stress (at the tip
of the bar stiffener, and well as tensile stress in the plate) below the threshold
limit, which is 0.5 Fy(at least for A-36 material). Then there is no need to worry
about fatigue failure of the metal

Weld failure is more critical, and is addressed by the AISC table and illustrative
examples. The weld that you have between the plate and the bar corresponds to
illustrative example no 4 (in the first edition, sorry we dont use AISC and my copy
is old, only for reference!). Basically, this is a double fillet weld between flange
and web of a beam. The stress in this weld is of the same nature as the weld between
the stiffener and plate that you have. The weld category is B. The loading condition
is 4. The stress you should be looking for is tensile stress. Here is the catch. If
you have the section designed as a T-section, the longitudinal stress in the weld
metal would be always zero - compressive under wheel load. Thus, theoretically,
fatigue failure doesn't take place. However, there are always very high locked up
tensile stresses in any welded region. Thus, under load, the actual stress will go
from more tensile to less tensile. You should therefore treat the stress range as
tensile and limit it to16 ksi. (The concept of taking stress range from max. tension
to max. compression on account of locked up stresses is a standard offshore industry
practice, where fatigue analysis of structures is routinely carried out). Thus, if
you size the plate + rib as a T-section with top stress limited to 16 ksi and bottom
stress to 0.5 Fy, you have taken care of fatigue.

Hope this helps,

M. Hariharan
Engineers India Limited

Bill Polhemus wrote:

> I am doing a fatigue analysis of a steel cover for a shallow vault set in the
> pavement of an interstate highway traffic lane. (Yes, it DOES sound like a dumb
> place for it, but there are extenuating circumstances that are not germane to my
> question, so just bear with me).
>
> The basic design of the cover is a steel plate with angle steel attached via
> fillet welds as "ribs" to the underside, with some rectangular steel bar
> material running around the perimeter as a "frame", also fillet welded.
>
> What I want is to run my approach by you, and get your comments. In particular,
> I need some comment on the design for fatigue, which is not commonly done for
> structures.
>
> First of all, I have applied as much of the AASHTO Bridge Design Spec as I
> possibly could, considering that this is a VERY special case. I applied a 75%
> impact factor for example, which mirrors what is required for bridge expansion
> joints, taking into account the commentary to the code that mentions this being
> appropriate for "discontinuities" in the pavement. Since the box is small, of
> course, I'm using a single "wheel footprint" of 10 inches by 20 inches, as
> AASHTO describes. I'm using a 16,000 lb. wheel load per the AASHTO standard
> truck.
>
> This gave me a pretty hefty requirement for design for strength, and now I need
> to look at this for fatigue.
>
> Now, the design for fatigue under AASHTO uses a pretty good-size load of 75% of
> the design truck load. However, I happen to know that this particular location
> will see pretty much NO truck traffic. It will, however, see quite a lot of
> transit bus traffic, and I have determined that a bus single-wheel load of
> 10,000 lbs. looks reasonable. I have decided to use this for fatigue, also
> adding a 75% impact factor.
>
> Since AASHTO is concentrated exclusively on trucks running on bridges, I've
> decided that for fatigue, it doesn't give me the design procedure I want to use.
> I have decided to use instead, the AISC procedure, which is more generalized,
> but which also looks like it uses the same stress "threshold" as AASHTO (I
> assume they are based on the same research).
>
> I have a fairly accurate accounting of the number of bus trips per day, and a
> reasonable estimate of bus trips per day. I am assuming that about one-third of
> the bus wheels that run over this road will hit the cover, and all that together
> tells me that I am looking at between 500,000 and 2,000,000 cycles over the
> lifetime of the vault cover.
>
> What I'm really wondering about, though, is how to apply the various "type and
> location" requirements that are shown in Table A-K3.2 of AISC LRFD. I am
> assuming that the critical elements are, of course, the "ribs", the bars making
> up the "frame", and the attendant fillet welds. However, since these are
> attached to the cover plate, it isn't clear to me how they need to be treated as
> they don't seem to correlate with any of the "illustrative examples" found in
> the AISC LRFD manual on pages 6-142 and -143.
>
> Anyone have any suggestions?
>