Latest / Mechanical Engineering Made Simple / (#153) The Design Junkie Vessel Survival

Transcript

(#153) The Design Junkie Vessel Survival · Mechanical Engineering Made Simple

0:00
Welcome back to mechanical Engineering made simple.
0:02
You asked for this pressure vessel series.
0:05
I delivered. I started exactly where most of
0:07
you are, machinist, welder, teaching myself after hours with
0:10
no degree and no handouts. Pressure vessels don't give a
0:13
fuck about paper on the wall, they only respect attention to
0:16
detail and reputation. Let's finish the series strong.
0:21
Before we get into the nuts and bolts of today's deep dive, I
0:23
want you to just picture what actually happens when a 5000
0:27
gallon pressure vessel fails on a factory floor.
0:30
Oh, it's, it's catastrophic. Right.
0:32
It doesn't just, you know, spring a leak and puddle on the
0:34
ground. It unzips, it turns into a
0:37
localized bomb, basically sending steel shrapnel through
0:40
brick walls. So today we are dissecting the
0:43
invisible physical laws that stop that from happening.
0:45
Yeah, the actual physics of it. Exactly, we are taking the
0:48
questions you brought us from the shop floor and following the
0:50
life cycle of a pressure vessel from raw steel and fit up
0:53
through the machining and welding all the way to the
0:56
proving grounds of certification and field erection.
1:00
Because I mean out on the shop floor you are constantly
1:03
negotiating with the physical reality of the metal long before
1:07
you even strike an arc. The underlying physics of the
1:10
material dictate absolutely everything about the vessel
1:14
survival, right? We just naturally tend to assume
1:16
that while making something thicker or using a higher
1:19
tensile strength alloy is always the ultimate answer, you know to
1:22
get containing pressure, but physical geometry plays by a
1:25
completely different set of rules.
1:27
I want to dig right into that actually, because I've seen
1:29
situations where a massive thick walled vessel gets put into
1:33
vacuum service. Maybe it's undergoing steam
1:35
cleaning. The steam condenses rapidly,
1:38
pulls a vacuum inside the tank and it just boom, crushes
1:41
inward. It just folds.
1:42
It seems crazy that 1/2 inch steel plate can't handle a
1:45
simple 14 LB atmospheric pressure differential.
1:49
It fails by sudden catastrophic collapse.
1:52
Yeah, it's known as elastic instability or buckling, and
1:56
this happens well before the material reaches anywhere near
1:59
its actual yield strength. Wait, really?
2:00
Before the yield. Oh, way before.
2:03
When a vessel is under internal pressure, the stress, they're
2:06
trying to stretch the metal out. It's like blowing up a balloon,
2:09
right? The material naturally pulls
2:11
itself into a uniform, stable shape.
2:14
Makes sense? But external pressure acts like
2:16
an invisible hand trying to crush a soda can.
2:20
An unblemished, perfectly cylindrical soda can holds a
2:23
surprising amount of weight if you stand on it perfectly
2:26
straight. But the second you introduce a
2:28
tiny geometric imperfection. Like just tapping the side of
2:32
the can with a pencil. Exactly.
2:34
You tap it and it buckles and collapses instantly.
2:36
So it's not about tensile strength at all, it's about the
2:39
fact that a slight out of roundness creates an immediate
2:42
weak point where all that external force aggressively
2:45
concentrates. Exactly.
2:46
For thin walled vessels under external pressure, the geometry
2:49
of the part is the critical survival factor, which, you
2:52
know, directly ties into how we actually join those geometric
2:55
shapes together. Every fabricator needs to
2:57
intuitively understand the concept of joint efficiency.
3:00
Right, the math penalty. Yeah, it is represented by the
3:02
Greek letter Ada in the engineering math, but out in the
3:05
real world it's a direct measurement of how much we trust
3:08
the Weld. I always thought joint
3:10
efficiency was just a mathematical penalty that
3:12
engineers throw in to be safe. Like if you have a perfectly
3:16
radiographed flawless Weld, your efficiency is 100%.
3:20
Right, a factor of 1. But if it's less inspected, they
3:22
drop it to 75 or 85%. But why?
3:26
What is actually happening inside that joint that forces us
3:29
to mathematically dock the vessel's strength?
3:32
Think about the physical reality of a Weld pool.
3:35
You are temporarily melting 2 pieces of steel into a liquid
3:39
state using intense electrical heat, often in dirty shop
3:43
conditions, and then letting it freeze back into a solid.
3:46
During that freezing process, invisible things go wrong.
3:49
Microscopic gas bubbles get trapped, causing porosity.
3:52
That's a flux. Or a slag get buried deep in the
3:55
root pass. Or a microscopic lack of fusion.
3:58
Exactly right at the side wall that an ultrasonic test might
4:01
miss if the angle isn't perfect. So the math is actually a safety
4:04
net for invisible physical defects.
4:07
Precisely. A joint efficiency of .85 means
4:10
the engineer mathematically has to treat the entire multi ton
4:14
vessel as if the steel is 15% weaker than its mill.
4:18
Test report claims it is. Yeah, that means the entire
4:21
vessel has to be designed with significantly thicker plate just
4:24
to compensate for the lack of absolute microscopic certainty
4:28
in the Weld. The care taking in the Bay
4:31
directly impacts the weight, the raw material cost and the
4:35
fundamental design. Which brings up the physical
4:37
assembly of those massive plates.
4:40
When a rolled shell plate is being fitted to a spondish head,
4:43
they almost never line up perfectly right out of the gate.
4:46
Never. The head might be slightly Oval,
4:48
the shell might have a flat spot from the rolling machine.
4:51
I see this out on the floor all the time.
4:52
Guys hooking up 10 ton come alongs and hydraulic jacks to
4:55
force the alignment. They pull the shell to match the
4:58
head and then lock it in place with a massive tack Weld.
5:01
Oh man, you can physically hear the metal groaning when that
5:04
happens. Yeah, the popping sounds.
5:06
Those popping sounds of those tack welds under extreme strain
5:10
should be a massive warning sign.
5:12
When you use brute force to bully the metal into a matching
5:15
geometry, you are causing what the codes call bulging.
5:19
It's exactly like cross threading a massive bolt on a
5:22
flange. Sure, you put a six foot cheater
5:24
bar on the wrench, you might get it to hold temporarily, and to
5:27
the naked eye the flanges look flush.
5:30
But you've destroyed the threads and baked in massive residual
5:33
tension. The cross threading visual is
5:35
perfect. You are locking in high
5:38
discontinuity stresses at the head to shell joint before the
5:41
vessel has even seen a single drop of operating pressure.
5:44
It's basically fighting itself. Right.
5:46
That joint is already fighting to tear itself apart just
5:48
sitting dead on the shop floor. When you add the actual internal
5:52
operating pressure later, those stresses compound.
5:55
The fabrication codes have incredibly strict fractional
5:58
tolerances for surface misalignment and out of
6:00
roundness for exactly this reason.
6:02
You cannot force a bad fit and expect the vessel to survive
6:06
long term fatigue. OK, so let's say the fit up is
6:08
done right, the geometry is natural, it's tacked cleanly,
6:12
and we are ready for the actual welding, machining and attaching
6:15
of components. We talked about how geometry
6:17
beats raw strength for external pressure, and I know the primary
6:21
way you improve that geometry is by adding heavy stiffening rings
6:24
to the outside of the shell. Yes, stiffeners are crucial.
6:27
But there is a massive rule here about where those rings actually
6:30
go. You must never, under any
6:32
circumstances, locate a stiffener directly over a
6:36
circumferential Weld seam. Which seems like a hassle when
6:39
you are trying to space out rings evenly for structural
6:42
support. Why is avoiding the Weld seem so
6:45
critical? It comes down to stress,
6:46
concentration and inspection first.
6:50
A Weld seam is already a geometric discontinuity.
6:53
The heat affected zone and the Weld metal itself have a
6:56
slightly different mechanical properties than the surrounding
6:59
parent plate. So it's already a weak point.
7:01
It is a natural place where stresses accumulate.
7:04
If you Weld a heavy stiffening ring directly over that seam,
7:07
you're taking a localized stress area and locking it down,
7:12
compounding the rigidity right where the shell needs to breathe
7:15
and flex. And what about the testing side
7:17
of it? Because if I cover up the main
7:18
seam joining 2 massive cans together, the quality control
7:22
inspectors are going to have a fit.
7:23
Oh, they'll reject it outright. Volumetric non destructive
7:26
testing becomes physically impossible if you need to shoot
7:30
X-rays or run an ultrasonic shear wave along that primary
7:33
circumferential joint to look for the microscopic porosity we
7:36
talked about earlier. A massive steel ring is blocking
7:40
the beam. You've blinded the inspector.
7:42
You've. Completely blinded them.
7:43
If spaced properly, stiffeners can absolutely double as
7:46
insulation support rings or platforms, but they must give
7:49
those primary seams wide clearance.
7:51
That makes total sense, but I need to push back on behalf of
7:55
the listener regarding the actual finishing of these welds
7:58
because this sparks huge arguments in the bays.
8:01
Welders take immense pride in their aesthetics.
8:03
I know the Instagram welds. Exactly.
8:06
A guy will lay down a bead and if it's slightly uneven he will
8:09
spend an hour with a grinder blending that seam until it
8:12
looks like glass. Are you seriously saying a
8:15
slightly ugly convex Weld that maintains its full thickness is
8:19
objectively safer than a perfectly smooth Overground
8:22
seam? I am.
8:24
It is absolutely true. Thickness equals survival.
8:27
Excessive grinding to blend a poorly aligned head to shell
8:30
seam, or just grinding to make a Weld look pretty is strictly
8:34
forbidden. Really.
8:36
Yeah, when you grind a well totally flush and you
8:38
inadvertently let the edge of that grinding wheel dip into the
8:41
parent metal alongside the Weld, you are physically carving away
8:45
the effective wall thickness of the pressure vessel.
8:47
You are literally grinding away the survival margin of the
8:50
steel. You are making the exact spot
8:52
that is most prone to failure thinner than the rest of the
8:54
tank. The ASME guidelines are brutally
8:57
clear on this. Deep grinding marks or grooves
9:00
deeper than 10% of the nominal plate thickness are entirely
9:04
unacceptable. Wow, 10%.
9:05
Yeah, a Weld profile that is a bit too convex might not win a
9:10
beauty contest on social media, but as long as it has smooth
9:13
transitions at the toes and doesn't create a sharp undercut
9:17
notch for stress to accumulate, it is vastly superior to a scene
9:22
that has been ground down below the nominal plate thickness.
9:25
That's a harsh truth for guys who want their work to look
9:28
flawless, but it's a necessary one.
9:30
Which leads us to what happens when we deliberately remove
9:33
strength from the vessel. When we cut a massive hole right
9:36
through a perfectly good pressure shell to attach a
9:38
nozzle or a small bore pipe, we have fundamentally weakened the
9:42
geometry. How do we mathematically heal
9:45
that wound in the steel? We restore that lost strength by
9:48
using a reinforcing pad, a doughnut of steel plate welded
9:51
directly around the nozzle. The critical rule here is
9:54
material matching. It is highly advisable to match
9:56
the pad materials tensile strength directly to the vessels
9:59
shell material. But out in the real world supply
10:02
chains get messed up. What if I need to reinforce a
10:04
high tensile carbon steel shell and the only plate I have on the
10:07
rack for a pad is a lower strength mild steel?
10:10
Well you have to do the math to compensate if a lower strength
10:13
material is used for the pad. You cannot just use the same
10:17
size pad you originally designed.
10:18
You have to upsize it. Right.
10:20
The area taken as reinforcement must be proportionately
10:23
increased based on the ratio of the stress intensity values of
10:27
the two materials. If your pad material is 20%
10:31
weaker than your shell material, you physically need 20% more
10:35
cross-sectional area of that pad to provide the equivalent
10:39
reinforcement. You can't just slap a piece of
10:41
weak steel on a high pressure shell and assume the hole is
10:44
healed. You have to respect the physical
10:46
forces. Exactly.
10:47
OK, so the fabrication is wrapped, the plates weren't
10:50
forced, the welds maintain full thickness, the pads are
10:53
mathematically sound. But we don't just fill it with
10:56
high pressure gas and call it a day.
10:58
It has to earn and its name plate.
11:00
Let's talk about the certification process starting
11:02
with post Weld heat treatment or PWHT.
11:05
We've all seen massive vessels being rolled into what look like
11:08
giant pizza ovens. Why do we bake the vessel before
11:11
we test it? We have to look at the atomic
11:14
level of the metal. Welding introduces intense,
11:17
highly localized heat upwards of thousands of degrees right at
11:20
the arc, while the steel inches away remains at room
11:24
temperature. This causes the crystalline
11:27
lattice structure of the metal to expand and contract violently
11:30
and unevenly. The atomic bonds get stretched
11:33
and twisted, trapping massive kinetic energy inside the heat
11:36
affected zone. We call these residual stresses.
11:39
So the metal is essentially acting like a stretched rubber
11:42
band that's been frozen in block of ice.
11:44
It wants to snap back, but the solid structure is holding it in
11:48
a state of constant tension. That is a great way to picture
11:50
it. And if you add internal
11:52
operating pressure on top of that frozen tension, the steel
11:54
shatters to melt the ice and relax the rubber band.
11:58
We use PWHT. We slowly heat the entire vessel
12:01
up to around 1100°F depending on the alloy and hold it there.
12:05
Kind of letting it breathe. Exactly.
12:07
The heat allows those distorted atomic bonds to slide back into
12:11
equilibrium, completely relaxing the steel.
12:14
For example, carbon steel spheres thicker than 1.5 inches
12:18
legally require this treatment, but the most critical rule is
12:21
that this stress relief must be fully 100% completed before any
12:27
hydrostatic testing begins. Because if I do the hydro test
12:30
1st and then throw it in the oven, I've completely changed
12:33
the mechanical properties of the vessel after its proving run.
12:36
I've invalidated the entire test.
12:37
Precisely. You can't prove a vessel and
12:39
then bake it. And Speaking of the hydro test,
12:42
there is a massive misconception what that test actually proves.
12:46
I used to think the hydro test was the ultimate guarantee.
12:49
You pump the vessel full of water, take the pressure up to
12:52
130% of the design limit and see if it drips.
12:55
If it doesn't pop, it's good for a 30 year lifespan, right?
12:58
Not even close. A hydrostatic pressure cast only
13:01
proves principal static stress on day one.
13:04
It proves the vessel won't burst under a heavy, unrelenting,
13:07
completely stationary load. OK, but vessels in the real
13:10
world aren't stationary. Right.
13:12
A pressure vessel in a refinery or a chemical plant doesn't just
13:15
sit there, it undergoes cyclic fatigue, the constant daily
13:19
expansion and contraction of starting up and shutting down.
13:22
Which flexes the metal back and forth millions of times.
13:26
And the hydro test does not prove the vessel can handle that
13:29
fatigue. It also doesn't test for thermal
13:32
shock. And most importantly, it does
13:34
not prove the vessel can handle creep.
13:36
Wait, creep. Let's talk about that because we
13:38
usually think of steel as an absolute rigid solid.
13:41
Well, at high temperatures, steel is not perfectly solid.
13:45
Over a decade operating at high heat and pressure, the metal
13:48
literally flows. It behaves almost like
13:51
incredibly cold honey. It's terrifying it.
13:53
Really is. It stretches and elongates,
13:55
imperceptible over time, That is creep.
13:57
The hydro test does not test for that.
13:59
A terrible Weld with poor internal penetration might
14:02
easily survive a static hydro test today because the
14:05
surrounding thick metal carries the static load.
14:07
But give it a few years. Yeah, three years from now,
14:09
under low cycle fatigue and creep, that same Weld will fail
14:13
catastrophically. The hydro test is a leak test,
14:15
not a guarantee of immortality. Let me ask you this too.
14:18
Why water? If we are testing a vessel that
14:21
will eventually hold compressed air or nitrogen, why do we fill
14:24
it with water for the test? Why not just pump it full of air
14:28
from the shop compressor? Because of the physics of
14:30
compressibility, water is virtually incompressible.
14:33
If you pump a vessel to 1000 PSI with water and a Weld fails, the
14:38
water instantly loses its pressure.
14:40
It just aggressively drips or streams out.
14:42
So the energy just drops. The energy dissipates
14:44
immediately, but air is highly compressible.
14:47
It acts like a giant coiled spring.
14:50
If a Weld fails at 1000 KSI of pneumatic pressure, all that
14:54
stored energy releases instantly.
14:57
Yeah, the air rapidly expands, tearing the steel apart and
15:00
turning the vessel into a fragmentation grenade.
15:03
Pneumatic testing is incredibly dangerous and is only used under
15:06
extreme, heavily controlled circumstances when a vessel
15:09
absolutely cannot support the weight of water.
15:11
That makes the reality of the stored energy so visceral.
15:15
So the vessel passes the hydro test safely with water, it
15:17
passes all the X-rays and the inspector finally stamps the
15:21
nameplate and attaches it to the shell.
15:23
This plate has the CE mark, the ASME stamp, the notified body
15:27
number. But there is one limit on there
15:28
that shop guys really need to pay attention to, right?
15:31
The MDMT. Yes, the minimum design metal
15:34
temperature. Every vessel and every
15:37
individual component welded to that vessel has an MDMT.
15:41
Which means what ractically? Well, carbon steel can become
15:44
dangerously brittle when it gets cold.
15:46
If it DROs below its MDMT, it loses its ductility.
15:49
Under pressure. It won't tear or stretch.
15:52
It will shatter like glass. To prove it won't shatter, the
15:55
steel has to undergo Sharpie V notch impact testing.
15:58
Where they hit it with a hammer. Basically they literally freeze
16:01
a small block of the steel, cut a notch in it, and hit it with a
16:04
heavy swinging pendulum to measure how much energy it
16:07
absorbs before it snaps. So the plate dictates the
16:10
absolute limits of the physics. But what happens if the vessel
16:13
is fully stamped, the nameplate is tacked on, and the foreman
16:17
realizes someone forgot to Weld on a tiny support clip for an
16:20
external pressure gauge? Can a welder just run over,
16:23
strike an arc and quickly tack it on before it ships out the
16:27
door? Absolutely not.
16:28
Once that name plate is attached, the vessel's identity
16:31
and certification are permanently locked.
16:34
No structural welding or modifications of any kind can be
16:38
made to the pressure boundary. None, even for a non load
16:41
bearing little clip. Zero.
16:43
The moment you strike an arc on that pressure boundary after the
16:46
nameplate is on, you're introducing new heat affected
16:49
zones, new residual stresses, and potentially microscopic
16:52
flaws that have not been accounted for radiographed or
16:55
hydro tested post certification. Structural changes completely
16:59
void the certification. It ceases to be illegal pressure
17:03
vessel. You just turned a multi
17:04
$1,000,000 engineered asset into a very heavy piece of scrap
17:08
metal because of a 10 second tack Weld.
17:10
That is wild. It happens more often than you
17:12
think. You have to respect the sequence
17:13
of operations. All right, so the vessel is
17:15
legal. It leaves the controlled
17:17
environment of the shop, gets loaded onto a flatbed, and is
17:20
erected out in the field, and now entirely new forces attack
17:23
it. We spent all this time worrying
17:25
about the internal pressure pushing outward, but the real
17:28
world brings localized loads. Right in the shop, general
17:31
pressure is uniform. The fluid or gas pushes evenly
17:35
against all the walls. But out in the field you have
17:38
eccentric supports like saddle legs, massive listing lugs and
17:43
trunnions. Plus the weather.
17:44
Exactly. You have environmental factors
17:46
like seismic activity or wind vibration causing vortex
17:50
shedding where the wind literally pulls the vessel
17:53
rhythmically side to side. These create massive local
17:56
loads. A lifting lug used to hoist 100
17:58
ton vertical tower induces severe compressive tensile and
18:02
shear forces right at the tiny footprint where it attaches to
18:05
the shell wall. It's like carrying an 80 LB
18:07
backpack by just one thin shoulder strap.
18:10
The overall weight of the bag is exactly the same, but because
18:13
that entire load is localized to one single point, it tears the
18:16
strap right off the fabric and destroys your shoulder.
18:19
That visual is exactly what happens to the steel in thin
18:23
walled vessels. These high local loads can
18:25
physically deform the wall of the vessel inward.
18:28
If the shell wall deforms inward, the lifting lug itself
18:31
tilts. And then everything goes wrong.
18:33
Everything that causes angular rotation of the lug, which
18:37
changes the angle of the rigging, which in turn can cause
18:39
the supporting crane cables to shift dramatically.
18:42
It is a terrifying cascading failure initiated by a localized
18:46
stress that wasn't properly reinforced with thick enough
18:48
insert plates or saddle rings. The stresses from these local
18:52
loads have to be mathematically added to the stresses from the
18:55
general internal pressure to find the true peak stress in the
18:58
metal. It all comes back to the math
19:00
and the physical reality of the metal.
19:02
You cannot outsmart it. Think about the sheer volume of
19:05
stored energy in a pressure vessel over a 30 year lifespan.
19:10
It is breathing, expanding, contracting, fighting off the
19:13
wind, the cold, the heat, creeping, imperceptible under
19:16
the strain. Every single day.
19:19
Your Weld is holding back an explosion.
19:21
That nameplate we talked about? It is a tombstone for bad
19:24
practices, but a monument to a craftsman's integrity.
19:27
It carries your legacy long after you retire.
19:30
Yeah, that's exactly what it is. This series is built for the
19:32
welders, grinders, and shop floor guys with no cushion.
19:36
I started the same way you are right now.
19:37
No degree, just hunger. If you're willing to outwork
19:40
everyone and protect your reputation like it's all you've
19:42
got, you will make it. Victory belongs to those who
19:45
refuse to lose. Ecial shout out to Air
19:48
Carbonark. This series is for you.
19:50
If you want a shout out and you need a topic covered that you're
19:53
struggling with, dro a comment. Thanks for riding with me,
19:55
design junkies. Tay hungry.