in your sketch orientation P2 would be a more efficient connection as it has the weld on the edge of the angle which is in tension for the P*e resisting moment couple. P1 the two vertical welds are being asked to handle P and the tension component of the resisting couple.

My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

To specifically address your question on capacity though:
If the welds are all the same effective throat and you ignore any bearing compression development with the embedded plate.

The two configurations have equal capacity using an elastic analysis because the elastic section modulus Sx is unchanged.

My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

The horizontal weld on P1 is more difficult to make.

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?
-Dik

Hey Celt83 - I don't think those two conditions will have the same Sx.

azcats - I believe that the S_x(bottom) of one configuration becomes the S_xtop of the other and vice-versa. I believe that is what Celt83 was alluding to in his identification that the Sx would be the same.

Azcats:
The only difference will be which is Sw,top or bottom with the same applied moment = P*e, consistent effective weld throat, and no compression development with the embed.

for Bending stress:
In the P2 orientation the max unit stress on the weld will be a compressive stress at the bottom of the vertical weld lines, P*e/Sw,bottom
In the P1 orientation the max unit stress on the weld will be a tensile stress on the top of the vertical weld lines, P*e/Sw,top

However with the flipped orientation Sw,bottom for P2 is = Sw,top for P1, so the magnitude of the stresses will be equal

For Vertical Shear Stress:
In all cases the contribution to the unit stress from P is just P/A,weld, where A,weld is assumed to equal to total length of weld.

With the magnitude of the weld unit stress being √ σ,shear^2 + σ,bending^2 , resulting in equal design unit stresses for either orientation.

My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

My mistake - I understand what you're saying now. My head was only looking at tensile stresses...

Blodgett's formulae derivation

• https://files.engineering.com/getfile.aspx?folder=279dcade-5fd7-4a43-8b57-5

and Hobert manual

• https://files.engineering.com/getfile.aspx?folder=82703098-a973-4369-bc0d-f

The fact is Ix is identical for both cases. Upon finding neutral axis, A_TOP = A_BOT, then for same bending on the cross section T ≡ C, and lever arm distance remains the same, so the capacity.

Quote (JohnRwals

P1 and P2 have exactly the same conditions except top, bottom horizontal welding.
I thought P2 could support greater vertical load.
But...
Which one do you think can support greater load? Why?)

I agree with you... P2 can support greater vertical load. If we consider shear , both alternatives have exactly the same shear capacity. However for bending, P1 upper sections of vertical weldments will experience more tension stress and will start yielding before P2 case. This is due to bearing compression development with the embedded plate which the NA will be lower than P2 case.

Are we all agreed that P2 will be greater than P1? I was a little confused by some of the answers, but the P2 arrangement is stronger because, while they are equal in shear capacity, for bending, P2 has the equivalent of 4 welds, top, bottom and 2 sides. No reason to ignore compression against the embed plate.

BA

Quote (Are we all agreed that P2 will be greater than P1?)

and to reiterate... it's easier to weld... seems like a win win situation.

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?
-Dik

Numerical example. Note the small discrepancy in tension and compression forces is due to roundoff.

r13, your calculation assumes the face of the angle against the embedded plate is extremely rigid compared to the embedded plate and you're neglecting any bearing pressure between the angle and the embedded plate.

For P1, the horizontal weld does nothing if the angle is placed firmly against the embedded plate - in this case there is direct bearing between angle and plate. If there is a gap then the weld must transmit the compressive forces, but it is more likely that the angle would be placed tight against the embedded plate.

For P2, I don't think the horizontal weld is fully effective until the vertical welds start to fail. Think about how the angle has to deform to deliver load to the horizontal weld...you would develop separation between the angle and embedded plate, which can't happen because of the vertical weld.

For both cases the horizontal weld is mostly for show. I consider both details to be equal. P2 would be better if you could flip the angle and put the vertical leg down and keep the weld on top.

R13, analyzing stresses in a weld group as equivalent couple forces is not a correct approach. In an extreme case, this method could overlook a peak stress at the tension toe of a fillet weld resulting in progressive/"unzipping" failure.

While we all appreciate your enthusiasm and contributions to this forum, you do tend to post incorrect or misleading statements with alarming frequency.

----
just call me Lo.

CANPRO,

The calculation was to clear up my comment made before - strength-wise, the two cases are identical.

I don't quite understand your bringing "bearing" into traditional weld design calculation. Question for you, don't you count the horizontal weld for shear, though it's not needed for compression?

I agree P2 with the angle flipped (or stiffened) is the highest capacity configuration.

I think that even as shown, the P2 horizontal leg will contribute slightly to the weld group capacity because of some weld ductility at the top corners. But depending on the proportions, I would be unlikely to rely on it fully or even partially due to the concerns CANPRO raised.

----
just call me Lo.

r13, draw a free body diagram of the angle - there are contact forces (bearing) between the angle and the embedded plate that most definitely effect the weld demand.

I would count the horizontal weld for direct shear resistance, but in comparing the two details this contribution is identical.

Quote (CANPRO)

P2 would be better if you could flip the angle and put the vertical leg down and keep the weld on top.

Quote (Lomarandil)

I agree P2 with the angle flipped (or stiffened) is the highest capacity configuration.

Do not agree with these statements if your going to rely on the compression development between the angle and the embed then flipping the angle so the flexible leg vs the rigid heel works to develop the compression block is an inherently worse condition.

My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

Lo,

I guess I understand the stress concentration/raiser problem at the weld corner. My mistake to use such example. But even with breaking up the weld lines, will both case stay the same? Please show me the "correct" method, thanks.

You think so Celt? I think your concern might be analogous to a concrete beam -- even if the leg is flexible and decreases the moment arm to the compression/bearing region, that's of less effect than the much larger area of weld (rebar) stressed in tension.

(For clarity, my comment is dealing entirely with the moment eccentric to the embed face. I'm not concerned here with the shear parallel to the face.)

----
just call me Lo.

Quote (CANPRO)

For P2, I don't think the horizontal weld is fully effective until the vertical welds start to fail. Think about how the angle has to deform to deliver load to the horizontal weld...you would develop separation between the angle and embedded plate, which can't happen because of the vertical weld.

By this same logic you wouldn't develop any compression bearing with the embed until the vertical welds start to fail. Under the assumptions of an elastic analysis the strain at the tip of the vertical welds is constant across the horizontal weld so the weld experiences stress.

Quote (CANPRO)

For P1, the horizontal weld does nothing if the angle is placed firmly against the embedded plate

agree, however in my experience more often than not the embed plate is askew and the angle seat is constructed with some small gaps filled with weld material. So so you'd need to work thru the weld deformation prior to engaging the plate in bearing, so the conservative design capacity would only consider the welds effective, then I have the bearing compression block as back pocket help if needed.

Quote (DIK)

and to reiterate... it's easier to weld... seems like a win win situation.

If we are talking about field welds absolutely agree, if were talking about a shop weld the weld to the angle toe probably requires more material as its flare bevel weld because the toe fillet vs standard fillet at the right angle heel. If this is Bent plate and not an angle then I am in full agreement with your statement.

My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

R13, your illustration shows all that is needed (I haven't checked your math).

For load in the downward direction, parts in initial contact, and a sufficiently stiff angle (which may or may not be present), the peak tensile stress is at the top of the weld group.

As you illustrate, because of the neutral axis location, the tensile stress (stress=M/S) at the top is lower for the "P2" configuration (your 0.023 when using a unit moment) than P1 (0.035)

As a general rule, I'm rarely concerned with the compressive stress in cases like this because of the possibility of bearing between the angle and embed plate (e.g. the weld isn't loaded). So I would check the combined tensile and shear stress at the top of the group and call it a day.

(Of course, some rare geometries and/or applications will change the validity of that last statement.)

----
just call me Lo.

Quote (Lo)

even if the leg is flexible and decreases the moment arm to the compression/bearing region, that's of less effect than the much larger area of weld (rebar) stressed in tension.

If you go with this analogy a smaller moment arm begets larger tension/compression couple forces required for equilibrium which results in higher tensile stresses in the weld.

My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

Quote (Lo)

R13, analyzing stresses in a weld group as equivalent couple forces is not a correct approach.

The method used was the classic elastic method for stress analysis, with the weld treated as a line. I don't know another method/approach, and would like to learn. Can you share, please.

Quote (lo)

As a general rule, I'm rarely concerned with the compressive stress in cases like this because of the possibility of bearing between the angle and embed plate (e.g. the weld isn't loaded). So I would check the combined tensile and shear stress at the top of the group and call it a day.

You are entitled to your approach, but it's something I wouldn't do - weld the tension side only. Is this statement sounds "incorrect" or "misleading", please advise.

The weld design will be done by 2 steps without calculating all that properties and stresses - weld for tension, and weld for shear. But this forum is not the place to advocate for design short cuts, unless it is a discussion between two members that fully aware what they are talking about. When you want to criticize somebody else's comment, it's fine, but please be clear, don't play opinionated/sentimental word games.

Suppose we eliminate the vertical welds in P1 and P2. Which carries more load?

BA

BAretired.

P2 would carry more load because P1 would probably be on the floor!

Quote (BA)

Suppose we eliminate the vertical welds in P1 and P2. Which carries more load?

P2 carries more because P1 will carry zero. P1 would pivot on the horizontal weld.

Celt83, I saw your reply to my comment...pretty busy so I didn't read over that closely, but I hope to later.

How about take away the horizontal weld for both case, which one is better?

Quote (EZBuilding)

P2 would carry more load because P1 would probably be on the floor!

Okay, so the horizontal weld in P2 is better placed than the one in P1; so why is this not still true when vertical welds are added?

BA

Quote (r13)

How about take away the horizontal weld for both case, which one is better?

They would be identical, r13, so the failure loads would be the same.

BA

Suppose we are connecting an HSS with point load at the free end to a rigid support and weld in the P1 or P2 pattern using a 3/8" fillet weld. Which would support more load assuming equal distance from support to load?

BA

BA,

I don't get that. The geometry center of the angle does not align with weld center then. Can you elaborate your view?

Okay. See below. (DRAWING EDITED)



BA

It boils down to if you want to consider bearing compression development between the beam (angle, HSS, etc.) and the embed. I'm of the opinion that the load-deformation response of a weld is such that if you have even a 1/16" of separation you need to rely on just the weld for the load transmission.

If you allow compression bearing development:
P2 will always result in the higher capacity

If you do not rely on compression development between the beam and embed and define failure as weld rupture:
P2 and P1 have equal design capacity

P2 is ultimately a safer condition because after you reach the failure criteria of the vertical weld lines in compression you then have the additional redundancy of the compression bearing development creating an additional load equilibrium condition.

My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

Quote (Celt83, BA)

It boils down to if you want to consider bearing compression development between the beam (angle, HSS, etc.) and the embed. I'm of the opinion that the load-deformation response of a weld is such that if you have even a 1/16" of separation you need to rely on just the weld for the load transmission. What can I say, other than I don't agree? No one is going to leave a gap between angle and embed plate.

If you allow compression bearing development:
P2 will always result in the higher capacity. Correct

If you do not rely on compression development between the beam and embed and define failure as weld rupture:
P2 and P1 have equal design capacity. Perhaps, but how many engineers would do that?

P2 is ultimately a safer condition because after you reach the failure criteria of the vertical weld lines in compression you then have the additional redundancy of the compression bearing development creating an additional load equilibrium condition. Indeed!!

BA

Quote (BAretired)

Perhaps, but how many engineers would do that?

I'm almost positive that every All-Welded Unstiffened Seat connection or Extended Shear Plate if pulled from the AISC tables have been designed on the assumption that the welds transfer the force to the underlying supporting member. Now the AISC detail for the Unstiffened seat only has a minimal return weld which I believe is because a full transverse weld ends up loaded in dominantly out of plane tension which if I'm remembering correctly has a pretty brittle failure mode (edit: I think my memory is off here I think its the vertical shear not the tension, since in theory the vertical shear load angle relative to the weld axis is 90 degrees which would let you take a multiplier on the weld capacity, but I think the end result wouldn't align with the ductility assumptions for the "pinned" connection. Need to reread the spec section on the topic)

So back tracking a bit the better connection detail would be to match the AISC recomendations.


My Personal Open Source Structural Applications:
https://github.com/buddyd16/Structural-Engineering

Open Source Structural GitHub Group:
https://github.com/open-struct-engineer

Celt83, Okay, you may be right. I am not conversant with the AISC tables.

BA

Celt83 hit the nail on the head in his second post on the 12th October if we are solely looking at the weld strength capacity then the two configurations would be the same. Fillet welds are designed based on shear stress whether it be shear stress from bending on the tension or compression side of the beam. The simple analysis carried out by treating welds as lines tends to be on the safe side and in addition a designer would use a safety factor on the stresses in any case.
So provided the shear stress is kept within allowable limits for the external load applied to the angle it doesn’t matter which way round the angles welded.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

The question was:

Quote (JohnRwals (OP))

Which one do you think can support greater load? Why?

Whatever AISC, Celt83 or desertfox says, I think the answer is P2.

EDIT: In fact if they were both built properly , P2 would be a great deal more than P1.


BA

BA,

From analytical point of view, all mentioned parties are correct. You are correct on your conviction, from material strength point of view, as there are literatures indicating that the transverse weld has higher strength than the vertical weld. Correct me, if I am wrong.

Quote (r13)

From analytical point of view, all mentioned parties are correct. You are correct on your conviction, from material strength point of view, as there are literatures indicating that the transverse weld has higher strength than the vertical weld. Correct me, if I am wrong.

You are not wrong about the strength of a weld at 0 or 90 degrees to the line of the weld, but there is the same amount of transverse weld in both schemes, so that does not account for my statement above.

The reason is simply that the top transverse weld has a moment arm equal to the height of the angle. The bottom weld does not. In both cases, failure will be by the angle rotating about the bottom. Scheme P1 will fail by tearing of the vertical weld at the top. Scheme P2 will fail when the top transverse weld pulls away from the embed plate while the bottom of the angle bears directly against the embed plate without the need of a weld. The difference between load P1 and P2 will be substantial.

BA

Wow! This is not a simple problem.
Thank you very much for diverse opinions!

I would like to use r13's elastic analysis approach which referred to the specific condition of M=P*e=1.
According to r13's analysis, as P (M) increases, failure will start where stress 0.035 is with M=1.
These failures happen at the end tips of vertical welding portion (farthest from horizontal welding);
P2 at the compression (bottom) area, and P1 at the tension (top) area.
In the case of P2, the bottom edge of angle can take compression after welding is crushed
if contact/bearing between angle and embed plate is possible as some engineers mentioned.
But, in the case of P1, as soon as failure starts at the end tip of welding, it will proceed like brittle failure.

Therefore, P1 and P2 have the same capacity as long as they are within the elastic limit.
But, beyond the elastic limit, P1 will cause brittle failure
while P2 can resist additional load as long as it can take advantage of bearing strength from embed plate.
So, P2 is better as some engineers' intuition pointed to...

Questions/Lessons:
Do you check the compression stress, compression failure of welding?
Many engineers used elastic analysis. What about inelastic/plastic method?
Don't they provide another insight?

Thanks!

JRW

Quote (JohRwals)

Therefore, P1 and P2 have the same capacity as long as they are within the elastic limit.

Even if that were true, that is not the capacity by definition. The capacity is the failure load and that does not occur while everything is elastic.

But beyond that, even within the elastic limit, the two schemes are not equivalent because in scheme P2, the bottom of the angle bears directly against the embed plate. That is equivalent to a full transverse weld in compression which cannot resist applied shear.

BA

BA,

We have to set the two conditions on a equal base. In general, we won't adopt P1 for many practical reasons, but if necessary (for no choice), would configuration of P1 support a lesser load than P2, assuming all other materials are adequately sized for the load. The failure mode addressed by you is one of the practical reason that we shy away from P1, but it does not affect the designed strength, if designed/calculated correctly. I think we are recognizing/responding the OP's question from different angles, though the conclusion is the same - The weld configuration of P1 and P2 will yield same computational strength, but P2 is preferred for practicalities.

Quote (r13)

the conclusion is the same - The weld configuration of P1 and P2 will yield same computational strength, but P2 is preferred for practicalities.

That seems to be what you are saying. It is not what I am saying. If properly constructed, P1 and P2 do not yield the same strength, computational or otherwise. Scheme P2 is preferred because it is stronger.

BA

weld stresses are complex but we have a simple analysis which allows us to calculate the average shear stresses in the welds and with given safety factors and a few assumptions the analysis errs on the safe side.
Now if you use this analysis it would show that the weld shearing stresses for the two configurations would be the same in magnitude and as previously stated if within the elastic limit of the materials involved, the use of either would suffice, why would one want to start analyzing what happens after yield, at that point the joints already failed at the design stage.
How then do we calculate the additional force or the additional strength that P2 will give us over P1?
Or put another way lets see the calculation that shows the magnitude of P2 and compare it with that of P1 bearing in mind we can calculate P1 from weld stress analysis of treating welds as lines.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

Desertfox,

I believe that the P2 configuration would carry about 11/7 or 1.57 times as much as the P1 configuration assuming a square arrangement of welds and e of 3/4 times angle leg length. Please see below.

BA

BA,

I guess you assume the entire section has reached yield as shown. But the neutral axis does not seem right.

I think the point is that with the load being out of the plane of the weld group, there is tension or tearing action in addition to vertical shear on the upper two corners of the weld in example P1. Given that this is where the weld starts or ends, there is a greater chance of a discontinuity that would propogate and cause a failure, particularly if the load was cycled or large enough to cause significant deformation of the angle. The weld group in P2 has much greater ability to resist this. In my opinion, P2 has a much greater chance of sucess over the life of the attachment. The overhead weld in P1 also makes it worse, in my opinion. Now, if the angle was a 6 x 6 x 1 and the force was 25 pounds then both weld groups are OK but I'd still go for the one that was easier in the field or shop.

BAretired

Cant see how you get Mmax for P2?
Looking in Mechanical engineers data by james Carvill the Z value for the weld treated as a line at position P2 is given by

Z= (2bd+d^2)t/3 at the top

Z= d^2(2b+d)t/(3(b+d)) at the bottom

Z = bending modulus, d = depth , b= width

So the bending modulus Z at the top and bottom are different as stated by others in previous posts, however it doesn't matter which way up you weld the the angle all that happens is the Z values top to bottom switch over but the overall capacity of the weld to take bending remains the same, which is exactly what Celt83 stated on the 12th October, or at least thats how I read it.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

For limit state, the area above and below the neutral axis are the same, so the ultimate tension and compression are identical, as well as the lever arm, for P1 and P2 both. However, P2 is superior, if we consider the additional capacity of the weld in direction transverse to load (horizontal weld), which was conservatively ignored by the code.

Here is a reference to analysing welds and the assumptions made:-

http://web.aeromech.usyd.edu.au/MECH4460/Course_Do...

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

I recall reading once that you should only consider the welded length as transferring load unless special effort is made to ensure contact for bearing. Eg the angle could be slightly warped so you don't get the bearing along the bottom edge, at least not initially. I couldn't easily find the document though to re-read all the comments.

desertfox:

I have included a snippet from your reference below. Please note that in North America, the symbol Z represents plastic modulus and the symbol S represents section modulus. I was (or thought I was) using plastic properties for my calculations. I'm not sure why that is wrong, but in any event, that is not the main source of our difference. In fact, using your formulas, the discrepancy is even larger.

Quote (desertfox, BAretired)

Cant see how you get Mmax for P2?
Looking in Mechanical engineers data by james Carvill the Z value for the weld treated as a line at position P2 is given by

Z= (2bd+d^2)t/3 at the top . . If b = d = 4 then Z_w = 16 (top of P2, ignoring direct bearing)

Z = bd + d²/3 top or bottom of P2, recognizing direct bearing . . Zw = 16 + 16/3 = 21.3

Z= d^2(2b+d)t/(3(b+d)) at the bottom . . Z_w = 8 (top of P1)

Z = bending modulus, d = depth , b= width

So the bending modulus Z at the top and bottom are different as stated by others in previous posts, however it doesn't matter which way up you weld the the angle all that happens is the Z values top to bottom switch over but the overall capacity of the weld to take bending remains the same, which is exactly what Celt83 stated on the 12th October, or at least thats how I read it.

Using your formulas and recognizing bearing at the bottom of the angle, the failure load of P2 in bending is 2.67 that of P1.
P1 = 8w/3 = 2.67w whereas P2 = 21.3w/3 = 7.11w.

It boils down to the question of whether or not you recognize direct bearing of the angle against the embed plate. If you do (and I do) then the P2 connection is much superior to P1.

Even if you don't recognize direct bearing, P1 is liable to unzip at the top of the vertical welds whereas P2, at worst will crush a bit of side weld, causing it to come into direct bearing with the embed plate.

BA

BAretired

The treatment of welds as lines doesn’t recognise direct bearing if you look at the last link I gave on the first page it clearly states that the weld only is analysed it also makes reference to Blodget and various other references including Shigley.
So by assuming that the weld only carries the load for either configuration is a more conservative calculation, further if there is no direct bearing as the last poster steveh49 alludes to then you could end up with at joint that fails.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

First time done limit state based weld calculation. Please advise for mistakes, if any.

No mistake, r13 other than the assumption of no direct bearing at the bottom. T1 alone provides a moment M = 4Fy*4 = 16Fy when direct bearing is considered. The question the OP asked was:

Quote (JohnRwals)

Which one do you think can support greater load? Why?

I think I have answered that question, notwithstanding opinions to the contrary.

BA

Quote (desertfox)

So by assuming that the weld only carries the load for either configuration is a more conservative calculation, further if there is no direct bearing as the last poster steveh49 alludes to then you could end up with at joint that fails.

In a way, it is not more conservative because it puts an extremely poor detail, P1, on a par with an extremely good detail, P2. Failure is far more likely to occur with the former than the latter.

BA

BAretired

I disagree totally with you there, if you calculate the Z values on either side of the neutral axis for configuration P1 or P2 the overall Z is the same however the lower value of Z is the one above or below the neutral axis without the horizontal weld, so you would size the weld on the lower value of Z, that in turn ensures that the weld is adequate on the weaker side of the joint and keeps the stresses well within the allowable. I would rather design on that basis than assume something you can’t guarantee (bearing compression)and run the risk of over loading a joint.
Doing it using the elastic method ensures that the joint is safe and if there was any bearing compression to be had, then that would be a bonus. It’s worth noting I have never seen any codes or specifications that take into account bearing compression in welded joints.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

To answer the question correctly it should be the one with the most conservative design, which in this case would be the elastic method and not assume benefit from a bearing compression which cannot be guaranteed.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

Fine then, desertfox. We agree to disagree.

BA

BA,

At the case for P1, wasn't there a force of C1 = 4fy (see sketch). With or without weld, the compression was counted. I am confused on your used of full member depth as the lever arm.

r13,

That is really what the disagreement has been about. It is not a theoretical disagreement as we all agree (I think) about the capacity of the weld when direct bearing is not considered. When direct bearing is considered, reactions T1 and C1 are separated by 4 units.

If the back surface of the angle or the exposed surface of the embed plate is not planar, the argument is that there could be bearing somewhere between b and c (see below). If that were the case, however, inspection should reveal a gap at point b, in which case steel packing could be added between the two surfaces. It is also true that, as P2 is increased, any gap at point b will tend to close.

BA

Got you. It is more complicate than I thought.

Hi r13

I have copied an extract from the Steel Construction Manual see attachment it clearly states the calculation doesn't take account of any compression connection between mating parts, which as BAretired said is were we differ.
Also I have pasted the link to the site for reference.
https://www.bgstructuralengineering.com/BGSCM14/BG...

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

• https://files.engineering.com/getfile.aspx?folder=52f411df-1a02-49fb-8673-5

desertfox,

We are in line with each other. The compression BA mentioned is something else, that is usually not considered in the weld design. Thanks for the reference, which is quite illuminating.

Quote (desertfox)

The forces are considered to be resisted by the weld group without taking into consideration any contribution from the compression between the connected parts.

I might do the same, especially now after all this discussion, but if you were to ask me which would carry more load, the weld as shown on your sketch or a weld placed on the bottom, I think you know what my response would be.

BA

I guess this is the case some people has in mind.

Not quite. Line stress on the bottom as well.



BA

there shouldn't be any line stress in the lower half the weld is only on three sides


“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

Quote (desertfox)

there shouldn't be any line stress in the lower half the weld is only on three sides

But cheer up, desertfox. If the point load is rotated 180^o, then P1 is the stronger configuration.

BA

The diagram I posted is illustrating the stress in the weld only, which is actually what I thought we were interested in for the elastic method.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

BA,

Can you show the axis of rotation?

desertfox,
The diagram I posted is the stress in the weld, assuming the heel point is fixed against horizontal movement.

r13,
The axis of rotation is the heel of the angle.

Quote (JohnRwals)

P1 and P2 have exactly the same conditions except top, bottom horizontal welding.
I thought P2 could support greater vertical load.
But...
Which one do you think can support greater load? Why?

I believe your instincts were correct. P2 can support greater vertical load than P1, provided that the heel bears directly against the embed plate.

If direct bearing at the heel is needed in order to safely sustain P2, it would be conservative but prudent to add weld across the heel, creating a closed rectangular pattern of welds as shown in the snippet from desertfox's reference.

BA

I think that there are two different questions here:

• Which connection has a higher design capacity
• Which connection has a higher actual capacity before failure

P2 almost certainly has a higher actual capacity before failure because there will be some amount of compression developed between the bottom of the angle and the plate (especially if you are defining failure as the angle falling down, not the first propagation of cracking in the welds). The problem is that you may not want to count on it for design for the variety of reasons brought up (gaps, compatibility issues, AISC doesn't do it that way). But given the choice, I would much rather use P2 since it does have some extra reserve.

Thank you, chris3eb. You have said it better than I. I interpreted the OP's question to be about capacity before failure.

BA

chris3eb,

I've not check the number yet, but both cases do have the component of compression, as long as the rotation is not about the bottom most edge of the angle. And, the compression is due to metal against metal, which turns out to be the bearing below the neutral axis/axis of rotation.

BAretired

So the stress distribution on your picture you have all the welds in tension except were the heel of the angle is, which apart from the fact that it is an assumption, you also claim it’s equivalent to a fillet weld in compression?
But according to AISC, Blodgett, Steel Construction Manual and several others using the elastic method, then the weld is subject to both compression and tensile stress.
Guess I won’t be writing to them to say they are incorrect.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

desertfox,

BA's approach is counting the welds above the neutral axis only. Thus the couple is the effects from tension on the welds, and the plate bearing (compression) from below. I think it has merit for further look, but not for design.

r13,

In P1, there is certainly no tension between the top of the angle and the plate. In P2, there may be some compression between the bottom of the angle and the plate - that is the extra capacity I am referring to.

desertfox,

I agree that the last diagram posted by BA is incorrect. Using the assumptions he previously mentioned, the vertical welds should be half in compression and half in tension. With his method, the angle would act the same as if it had four welds when resisting moment (but only 3 resisting shear). I think most are in agreement that the design capacity would be the same - BA mentioned that a couple of days ago. If these angles were load tested to failure, do you think that P2 would hold more load? If not, is your thought that by the time any gap closed up, the welds would already be past the point of failure?

desertfox, r13:

I was simplifying it as shown below. All welds above the hinge would be in tension.



BA

This is what I was describing. There is a compression component in each case.

Hi chris3eb

In truth if we are looking at destroying the joint the only way to find the ultimate load would be a mechanical test and I don’t doubt that configuration P2 would probably be the stronger.
That said welded joints are complicated to analyse but the joint as to be designed in the first place and there may be lots of reasons why configuration P1 might be required. Using the elastic method as previously stated would mean either configuration designed for a given load would work with no issues. Even if P1 cannot benefit from an assumed bearing compression it on doesn’t make it a poor configuration if designed correctly.
What I also said is that it is better to use the elastic method and size the weld on the weaker Z value, therefore ensuring no working failure on the joint but if that joint configuration can benefit from some (assumed)bearing compression then that would be an added bonus. The problem I foresee designing it on what BAretired posted earlier is that the weld size would be under estimated and thereby fail in service.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

Maybe.

We really have become bogged down in a lot of garbage brought on by a nonsensical stipulation that we cannot bear metal against metal without having a weld. We do it all the time with beams on bearing plates. The sensible thing to do here, instead of adopting the 'U' shaped weld configuration is to place one weld on top of the angle designed to resist P*e/d and allow the bottom of the angle to bear, without a weld against the embed plate. For those who are afraid we may have a gap, feel free to add a thin strip of steel between the angle and the plate. In my view, that would be a waste of time and effort.

A weld across the top of the angle will have a unit tension of Pe/bd and a unit shear of P/b. One simple weld should do it.

BA

BAretired

It’s not nonsensical that we cannot assume no bearing compression, all the references for weld design don’t consider it. If of course you have some information that we aren’t aware of that allows the calculation to be done with bearing compression please share it. Even chris3eb agrees your stresses on the weld all being in tension are incorrect. If you do this calculation all the time then show us a good example with values of stress and we can all benefit.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

desertfox,

I too disagree with the assertion that the stresses in the vertical welds will be all tension; because the heel will move slightly as the angle compresses against the embed plate, causing a small compression at the bottom of the vertical welds. But it is close to being true. Neither your assumption nor my assumption is perfect but there isn't enough time to perform a finite element analysis on every angle welded to an embed plate. I believe my assumption is better than yours, but I would agree that, when in doubt, be a little conservative. With the inverted 'U' configuration, I would design the top weld to resist Pe/d on its own, without relying on the vertical welds.

In my last post, I suggested confining the weld to the top of the vertical leg of the angle and allowing the bottom to bear against the embed plate. The bottom support is simply a horizontal reaction. The lateral movement is trivial and does not affect the force carried by the top weld.

BA

BAretired

No principles that we use are perfect and of course that’s the reason for the safety factors and I agree that neither of us are perfectly correct. My only standpoint was that the joint needs to be designed in the first instance and all the researching I have done appears for the purpose of design, to use the elastic analysis and assume that all the force is transferred via the weld to the adjoining metal. I have attempted to calculate the stresses in the welds assuming that the U section weld in the configuration of P2 is rotating about its bottom edge where there is no weld.
All the welds are in tension and the maximum shear stress occurs across the horizontal weld at the top but sadly the magnitude of the stress is three times smaller than that found using the elastic analysis. I don’t doubt that there is a benefit from bearing compression and that configuration P2 is the more likely to benefit from it but I would prefer not to assume it’s there in the design but get it as a bonus.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

Looking back at the example in Steel Construction Manual, the applied force P appears to be at an angle to the vertical. I assumed it was intended to be vertical, but I could be mistaken. In any case, to ignore the horizontal force at the bottom centre of the plate is certainly conservative, but it is wrong. To determine weld stresses without taking that force into account is pure fiction.

BA

Yes in that picture the force was meant to be angled and I think if you follow the example through it accounts for the rotation of the bracket too, it was just an example I found which was close to what we were discussing.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

BA,

If I didn't misunderstand, this is what you are advocating - for weld design, the geometric property of the angle is the same as the channel, or a tube, with the same member width and depth, due to contribution of the bearing; so the weld size should be determined accordingly. I wouldn't say it is correct or wrong, but it complicates the weld design process, for loading conditions that are largely dependent upon the geometry center/location of the neutral axis.




If the sketches do not catch your idea correctly, you should provide an numerical example, that reflects your concept, and people can understand correctly.

Modified sketch.

For me, from a design perspective, P1 actually has zero capacity as shown below. That said, I would normally specify welds both top and bottom.

r13,

What I am advocating is a single weld at the top of the vertical leg of the angle...period. I showed it before, but repeat it here; the fillet weld is shown as a little green triangle at point c.

This is the simplest arrangement and makes maximum use of weld metal. Horizontal weld shear is T1 = P*e/d. Vertical weld shear is P. The horizontal reaction at point b is C1 = T1 = P*e/d where d is the vertical leg length (less a negligible amount to provide bearing).

Thin shims could be tacked to the embed plate at point 'b' to ensure the angle did not hang up on accidental projections, but I don't believe that would be necessary.

BA

KootK,

I couldn't agree more!

BA

BA's case reveals the inefficiency of the weld arrangement of P1 case offered by OP, but does not overturn the weld strength calculations based on literatures. It is too far to say P1 has zero capacity though, unless vertical welds are not considered.

According to the elastic method of weld analysis, it’s the two vertical welds where there is no horizontal weld, that will fail first irrespective of whether P1 or P2 is used.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

Sounds like a good argument for not having vertical welds, but with P1 and P2, the vertical welds are in tension and compression respectively at their maximum value. I would have thought the weld is less likely to fail when in compression.

BA

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

When member in full bearing (no gap between contact surfaces), the compression side weld does not contribute any strength. But for stress analysis, its property must be accounted for.

I’ve picked some figures and loads randomly for the purpose of illustration in the post above.
If the weld is turned to position P2 then the Z values are just reversed. The neutral axis for the weld outline is located 33.3mm below the horizontal weld, so the vertical legs always have the lower value of Z whether they point up or down, so for any given loading the will always carry the higher stress.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

Quote (desertfox)

The neutral axis for the weld outline is located 33.3mm below the horizontal weld, so the vertical legs always have the lower value of Z whether they point up or down, so for any given loading the will always carry the higher stress.

The maximum weld stress always occurs at the open end of the vertical weld, but with P1, it is preventing the angle from separating from the embed plate whereas with P2, it is preventing the angle from closing an imagined gap between itself and the embed plate. An unzipping, it seems to me, is more likely to occur with P1 than with P2.

BA

I wouldn’t disagree with you on that BA but as I stated way back provided we size the weld based on the lower Z value then for a given loading,P1 would be okay but it’s not the best use of the material.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

I am afraid the thread will disappear together with more than a hundred responds . May be OP will delete and go away ?.

I commented to thread in another forum. The thread disappeared , which i am thinking the OP deleted and went a way..