soc.history.war.world-war-ii

This is taken from the book Liberty by Peter Elphick, the chapter
entitled "close to calamity" which deals with the emergence of
structural failures in US built WWII merchant ships and the
investigations undertaken to solve the problem.

As of 1 September 1944 some 558 ships had reported a total of
785 incidents of cracking, and this is known to be an underestimate
of the total.

The US "Final Report", containing data to 1 April 1946 noted of
some 5,000 Maritime Commission built ships some 972 had reported
a total of 1,442 incidents. Top of the list were a tanker and two
non liberty ships which had suffered 8 incidents, the "top" two liberty
ships had suffered 6 incidents.

One of the first incidents was when building the ship Ocean Justice, which
was to a UK order and not being built under the Maritime Commission,
a welder was amazed to watch cracks propagating, on a cold morning
in February 1942, the damage was classified as serious and the ship
was not launched for another two months. Of the 60 Ocean class
some 9 would suffer some sort of failure. As an aside the Ocean
class shows the US Maritime Commission report does not report all
US merchant ship construction, the Ocean class would add about 1%
to the total number of ships built by the US.

The next major incident reported was on 10 November 1942 and
the third on the 11th. By 12 January 1943 another 10 ships had
reported major cracking problems. The US assigned the category
class I to these, with class II being bad with the potential to become
class I and class III for other reports. There were 18 class II and III
incidents between November and January.

The Germans were well aware of the problem, even publishing an
account in the journal Nauticus in 1944. The US allowed the New
York Times to carry an article on the topic in April 1944. There is
little doubt a problem in sorting out what was going on were the
rumours, some of which were axis fuelled, as you would expect.

By the end of 1942 the cracking defects were clearly recognised as
a problem and on 16 January 1943 the new tanker Schenectady
nearly broke in half while being fitted out on the Columbia River.
The break sounded like an explosion. Air temperature was 26
degrees Fahrenheit. The cracking had occurred so quickly the
raised sections of the hull had almost no chance to ship water.
(The ship broke into an inverted letter V as the bow and stern
were carrying ballast).

It was clear the crack did not follow the welds but went through
the steel plates. The initial reaction was to blame the welds but
that was not sustainable. The steel was taken for experiments
and passed all the standard tests. The key observation was the
steel had behaved as if it were brittle, with the pieces able to be
reassembled into their original shape, whereas it should have
been ductile, with some ability to "bend and stretch" before
breaking and the distortions before breaking would mean the
pieces could not be reassembled so neatly.

Since a majority of the early incidents came from Kaiser yards it
was easy to assume it was a problem for that builder alone but
as the incidents continued from February 1943 onwards it was
clear there was a more systematic problem. It was across all
yards and all types of ships.

The next 11 class I failures resulted in the loss of two ships at
sea. Then the Esso Manhattan, a tanker, effectively split in two
off New York on 29 March 1943, with its two escorting ASW
blimps granted ringside seats to take lots of photographs. In
this case the crack had started at a defective weld. The ship
made harbour and was put back together.

The US started an official inquiry on 20 April 1943 with an
"utmost urgency" brief, to find out what was going on. At about
the same time the British initiated a similar investigation, both
countries gave the investigations considerable authority and
resources, and exchange visits were organised to keep each
investigation up to date with the other.

The US started by circulating to masters of US ships a survey
form requesting reports on cracking problems, accepting the
inevitable rumours that fuelled. The available data collected
revealed by 30 April 1943 there had been 28 class I events and
over 100 events in total. Later the British circulated a similar
survey. These surveys were continued throughout the war but they
inevitably missed some incidents. It seems the reporting system
stayed in use until the late 1960's.

Looking at the official US final report, which was data collected to
17 March 1946, there had been 132 class I failures. Peak class I
failure month had been January 1944, with 20, worst month for all
types of failures was March 1944 with 138.

It would appear it was not until November 1943 that a failure resulted
in US lives being lost, and that was from a lifeboat going missing.

Some 8 US and 6 British ships were used as tests, taking ships
out of commission as required.

The British tests were more elaborate than the US ones, using
still water trials of sister ships, one welded and one riveted, and
ocean trials with a ship whose master was instructed to go
through storms as much as was safely possible. Incidentally
these full size trials run in the 1944 to 1947 period were published
by the Admiralty and remain to this day the most modern set of
data on the stresses full size merchant ships encounter. Elphick
hints that maybe it is time for another set of trials. It took the
sea trials to finally convince the die hards that it was not a case
of riveting being superior as a joining technique.

The first obstacle any investigation had to overcome was the fact
ship building rules were not codified, they were a collection of rules
of thumb developed over the years. So no one knew what sort of
stresses a "normal", that is riveted, merchant ship would see.
This enabled those who preferred riveted construction to hold the
view it was welding causing the problem, ideas like faulty welds
or because riveting made the ship overall less rigid and so able to
absorb stresses better. It took the full size trials to disprove the
claimed superiority of riveting when it came to handling stress.
In simple terms the difference between riveting and welding was
minor. The crack resistance superiority came from the fact the
welds created one metal sheet, riveting kept the sheets separate.

As another example of the time needed it took about a year, that
is until sometime in 1944, before any researcher reproduced the
cracking in a laboratory experiment.

One of the actions first approved was the fitting of crack arresters,
4 long slots were cut in the ship's plating and covered by a strap
that was riveted into place, it was easy to do this for new ships
but expensive for ships already built, about 1,000 already built
Liberty ships received the modifications. Apart from structural
failures the crack arresters helped limit damage from bomb etc. hits.
Ships meant as troop transports were given additional strengthening.

The first sets of data showed as expected cracks started from parts
of the ship under high stress and strengthening was ordered for
key areas. Welding remained a favourite candidate for the ultimate
cause and more care was taken to ensure welding was to proper
standards and the crack down did uncover problems, where welds
were simply done incorrectly. A practice of poor welding was
wide spread enough to have its own word, slugging, to describe one
technique. Yet tests on slugged welds showed if anything they were
superior to the "correct" welds, as opposed to other cases where
the weld was simply faulty.

Another clue was British built welded ships were not suffering the
same defect rate.

It was also clear from the incident reports the cracks did not normally
follow the welds, indeed they often seemed to quickly move away from
them. Extensive US tests showed welds did not create "locked up"
or residual tensions which then created failure points.

The way the plates were cracking suggested the steel was the problem
and like welding more effort was put in to ensure quality control was
improved. As expected some quality problems were found in the steel
supply but again standard testing on cracked plates kept coming up with
pass marks. The Bethlehem Fairfield yard used part riveting in its ships
and they suffered a much lower rate of cracking, which was another point
in favour of the riveting was better belief. There was also the suspicion
Bethlehem was managing to make sure its best steel went to its own
ship yard.

The original April 1943 survey was also clearly pointing out low
temperatures were a problem. As an aside whoever produced the
survey questions managed to do a very good job, covering almost
all the information needed to solve the problem.

As the incident numbers built it became clear that ships built in
lower temperature environments were more at risk. It also
seemed to be ships in ballast were at more risk, and ballasting
instructions were changed

Another complicating factor was ships in US service had a higher
risk of seeing structural problems, ships in UK or Norwegian
service a lower risk, but in Soviet service the risk was higher than
for the US. While the Soviets were operating in lower temperature
environments Elphick points to the average US merchant marine
officer had much less sea experience than the British and Norwegians
and this may have lead to the US ships being subjected to greater
storm stresses as seamanship is more experience than theory.

In the end the metallurgists came up with the answer. The Cambridge
(UK) team was lead by Dr Constance Tipper. The term notch
brittleness is used to describe a plate having a higher stress area, a
notch, and when it would break in a brittle manner at low temperatures.
No tests were done for this at the time and it turned out US steels with
their lower manganese to carbon ratios showed the problem more than
UK steels. In almost all cases it required temperatures so low that
they were rarely encountered however some of the US steel batches
were so sensitive they could crack in tropical waters.

It explained why welding was not a problem, why UK welded ships
were not giving the same amount of problems, why it was worst in
winter and why temperatures when under construction seemed to
matter.

It would appear, despite the problems, the US wartime built ships
lasted in service in bulk until the 1960's when their age and the
superiority of the new diesel driven ships meant the wartime ships
were uneconomic. In one sense they had to last, as of 1945 the
US merchant fleet was bigger than the rest of the world combined.

Geoffrey Sinclair
Remove the nb for email.