Hoping someone can give some input on this issue.  We are having fabricated a 316 stainless steel shelf angle.  The angle is 8" high, 10" in depth and approx. 85" in length.  The material thickness is 3/4".  The top shelf, 10" piece, is water jet cut, then machined to +/- .005 of the required finish size.  After machining, the plates are welded together at a 90 degree angle with a full pen weld.  After welding, the machined plate is .175" shorter on the welded side compared to the non welded side.  Overall bow after welding is less than 80 thousandths.  Has anyone seen this happen before?  This is a first for us.  Thanks.

There are four states of matter: solid, liquid, gas, and plasma.

When pure metal is in the liquid state and it is allowed to cool the metal changes from a liquid into a solid. This is a change in state. If you are heating the metal from a solid and it turns into a liquid the magic temperature is the melting point. If you are cooling the liquid until it solidifies it is known as the solidification temperature. For a pure metal the melting point and solidification temperature are one and the same. Beginning with a liquid, the temperature drops steadily and then it plateaus or stabilizes and the temperature no longer drops. The temperature stabilizes (called temperature arrest) as the latent heat of fusion is dissipated by conduction or convection. Once the energy, i.e., latent heat of fusion, is lost, there is a phase change from a liquid to a solid. Upon heating, the reverse is true, i.e., it is a reversible process.

The phase change from liquid to solid is accompanied by a reduction in volume. The volumetric change resulting from the change in state is call shrinkage. It is on the order of 3% for iron and steel alloys. It is on the order of 6% for aluminum and aluminum alloys. 

If you have an alloy system the metal neither melts nor solidifies at a distinct temperature, but rather solidifies or completely melts over a temperature range. The temperature at which the metal is completely liquid is called the liquidus temperature. The temperature at which the alloy is completely solid is called the solidus temperature. In between the liquidus and solidus temperatures you have a mixture of solid and liquid. As the temperature approaches the liquidus temperature the liquid fraction increases. As the temperature drops and approaches the solidus temperature the solid fraction increases. An analogy would be slush in the winter; it is a mixture of liquid water and solid water in the form of snow and ice.

If you have a pure metal in the liquid state and you allow it to cool slowly without any vibration, you can experience a phenomena called super-cooling. An experiment can be performed using water that demonstrates super-cooling. Water typically solidifies at 32 degrees F, but slow cooling without vibration will allow the liquid to cool well below 32 degrees F. If you tap the container it will instantly turn to ice. I always have bottled water in my car. On winter mornings when I get into the car the bottled water is in the liquid state even if the temperature is down in the 20’s. If I give the bottle a shake, it turns to ice almost instantaneously. Try it, it still cold enough in many places to try this experiment. 

Back to the original conversation; Again, shrinkage is due to phase change, i.e., going from a liquid to solid.

Contraction is the dimension change associated with changes in temperature without a phase change. So, if you heat a bar of iron or steel alloy, it changes dimensions. The bar gets longer, the bar increases in diameter, thus there is a change in volume, but there is no change in phase. The dimensional change is linear. That is, there are changes in dimensions that are proportional to temperature.

The coefficient of expansion for iron is 6.7 X10-6th power or 0.0000065 inch per inch degree F. Ce for carbon steel is 7.3 x 10-6th and 304 stainless is 9.6 x 10-6th, while aluminum is 12.3 x 10-6th. If you heat the bar it gets bigger and the amount of increase can be calculated pretty precisely by multiplying the coefficient of expansion by the original dimension by the change in temperature. The relationship, i.e., coefficient of expansion is the same as long as there is no phase transformation. So, for practical use, you can use the coefficient of expansion for temperatures up to 1335 degrees F for carbon steel. That is the temperature where iron starts to transform from ferrite or pearlite to austenite. Then the coefficient changes to a different value. We all know that the dimensions gets larger upon heating, and small upon cooling. The process is reversible.

Now, when dealing with carbon steels, that temperature, 1335 degrees F, has special meaning. At that temperature the body centered cubic structure starts to transform into a face centered cubic structure. At temperatures below 1335 degrees F the iron is in the form of a body centered cubic unit cell. The BCC has 9 atoms in the unit cell (in the box), whereas the FCC has 14 atoms in the unit cell (in the box). When you start with 200 oranges (iron atoms) and you pack 9 oranges (iron atoms) in each box, you need roughly 22 boxes to hold all the oranges (iron atoms). If you repack the boxes and place 14 oranges (iron atoms) in each box, you only need 14 boxes. Clearly the total volume decreases by 8 boxes. That is a significant change in volume due to the change in phase, i.e., BCC to FCC or FCC to BCC.

Summing up:
Shrinkage – a volumetric reduction associated the change in state when the iron changes from a liquid to a solid.
Contraction – a change in volume due to a change in temperature (but does not involve a phase change).

Here’s a definition for shrinkage:
McGraw-Hill Science & Technology Dictionary
(metallurgy) Volume contraction of a metal during solidification.

I can’t quote a source or which book I took this from. I’m on the road in Toronto this week, so I don’t have access to my books. Sorry.

I don’t know if this helps or muddies the water even more.

Best regards - Al

There are a couple of things to consider;

1) if there is a dimensional change, there is also a volumetric change.

2) permanent distortion occurs when the applied stress exceeds the yield point of the base metal.

3) the stress resulting from shrinkage is very small because the strength of the metal as it undergoes solidification is nil. However, as the metal continues to cool the strength increases and the material contracts with a force equal to the yield strength of the metal (at the specific temperature). The strength  increases rapidly as the temperature approaches ambient temperature and is equal to the yield strength of the base metal once cooled to room temperature.

4) any base metal heated to a delta T of about 220 degrees F has reached the yield point, thus the residual stress for the entire region heated to the delta T must be considered. The majority of the residual stress is from the surrounding base metal heated to a temperature above 300 degrees F (ambient plus the delta T). The volume of base metal causing the residual stress and resulting distortion far outweighs the residual stress and distortion of the weld alone.
In the final analysis, the weld only contributes about 10% of the total distortion. That is why "peening" is of such limited benefit.

Thin plates experience less angular distortion than thick plates because the thermal gradient is less in the thin plate. This plates, welded with multiple passes experiences a higher thermal gradient in the through thickness direction, thus experience more angular distortion. The thermal gradient, the difference in temperature from point A to point be is responsible for he majority of the distortion. With that in mind, efforts to reduce the thermal gradient will reduce the total distortion. Preheating is an effective means of reducing thermal gradient and final distortion.

Fun stuff this distortion thing.

Flame straightening can be an effective aid to reduce existing distortion, but only if you maximize the thermal gradient. That is, you only heat a localized area, not the entire member.

Best regards - Al