Expert Witness Journal Issue 68 August 2026 - Flipbook - Page 31
Physics Doesn’t Care Who’s Paying
by Jason Madge, CEng MIMechE PhD & Lorenzo Baietta, CEng MIMechE at Elemeca.
Secondary damage is the most familiar obstacle.
When a gas turbine blade fails, the initial fracture
event releases energy that immediately begins
destroying evidence of its own cause. Downstream
components are struck by liberated material;
the fracture surface may be obliterated within
fractions of a second; witness marks that might have
distinguished fatigue initiating at a manufacturing
discontinuity from fatigue initiating at a thermally
degraded surface are gone before the investigation
begins. What the investigator receives is not the
failure — it is the aftermath of the failure.
The legal starting point
Expert witnesses in civil proceedings owe an
overriding duty to the court, not to the party
instructing them. Practice Direction 35 is explicit,
and the courts are increasingly willing to enforce
it with serious consequences. That much is familiar
to every solicitor engaged in technical litigation.
What receives less attention is a complementary
observation: the core requirements that the
procedural rules impose on engineering experts are
not foreign impositions on engineering practice.
On the contrary, this is what good engineering
practice demands. However, the two frameworks are
challenged not by a 昀氀aw in either — it is a human
one.
Some secondary damage is often inevitable – e.g.
due to continued operation or shut down after
the primary failure, is a reality. However, some of
it can be avoided: Careful storage and recording
of evidence and considered, strategic activity
planning to minimise destruction of other evidence
are powerful methods to build as full a picture as
possible.
Physics has no client
A fracture surface does not know which party
retained the expert who examined it. A fatigue
crack propagates under the same stress intensity
conditions regardless of whether the failed
component belongs to the claimant’s equipment or
the defendant’s product. Material properties, load
history, geometry — these are objective facts, and
they point in whatever direction the evidence leads.
Incomplete documentary evidence compounds
this. Maintenance records may have gaps.
Operational data may have been overwritten.
Borescope images taken prior to an incident may be
genuinely ambiguous — dimensional measurements
serviceable on one reading of the applicable
standard and unserviceable on another. The result
is that multiple failure hypotheses are commonly
supportable from the same physical dataset. Each
may be technically defensible. Each may imply a
di昀昀erent liability picture.
The engineer who follows the physics is already
doing what Part 35 requires. The engineer who
starts with a conclusion and works backwards is not
simply failing a legal obligation — they are doing
bad engineering. These two failures are the same
failure, arrived at by the same route.
Investigations must often take an iterative approach:
the hypotheses come 昀椀rst, and then each hypothesis
is tested against the evidence available. Some can
then be ruled out con昀椀dently. Of the remaining
theories the investigator must ask what information
could prove or disprove the hypothesis, and this
activity can be integrated into the evidence gathering
plan. This loop will often need to be revisited several
What root cause analysis actually
looks like
This would be straightforward if physical evidence
produced unambiguous answers. It rarely does. A
competent engineering investigation almost always
begins with a picture that is partial, contradictory,
and consistent with more than one explanation.
EXPERT WITNESS JOURNAL
29
AUGUST 2026
