Professional Documents
Culture Documents
DOI: 10.1061/(ASCE)CF.1943-5509.0000274 ment to the implementation of the forensic process is presented, and
practical steps are discussed that may assist in overcoming this
difficulty.
Introduction
sign process creates an appropriate design solution through ture would have behaved based on the design process.8 Finally, the
the application of simplifying assumptions, and, where appro- separation of evidence collection from the development of hypoth-
priate, errs on the side of conservatism. In new or remedial eses, in conjunction with the rigorous testing of each hypothesis
design, this is one of the design process’s chief strengths, but against the evidence, assists the investigator to conduct the inves-
in failure investigation, it is a critical weakness. In failure in- tigation in a forensically sound manner, ensuring it will not only
vestigation, the investigator must determine the actual loads, stand up to the scrutiny of engineering peers, but also, if necessary,
actual structural behavior, and actual material properties at to the exacting demands of the legal system.
the time of failure, rather than relying on simplifying perfor-
mance assumptions. This issue can be further exacerbated by
the sometimes-significant differences between simplifying Implementation of Forensic Process
performance assumptions and the performance of structures
in practice.7 Therefore, the accurate determination of the cause Given the significant differences between the forensic and design
of failure depends on verifiable evidence (e.g., a bolt’s processes, it is clear that Carper can make the point that “a good
failure surface or the cracking patterns in concrete members), design professional is not necessarily a good forensic expert”
and while the collection and analysis of verifiable evidence is (2000). The accurate determination of causation will be very diffi-
central to failure investigation, it is not an integral part of the cult unless the design engineer is able to put aside the traditional
design process. design process and apply a forensic process. In practice, however,
These limitations affect how an engineer that typically utilizes the ability of a design engineer to make such a transition may be
the design process approaches causation investigations. Although highly questionable. Fundamentally, without experience in foren-
determining causation is a critical objective, the implicit nature of sics, the embedded nature of the design process in an engineer’s
the design process can naturally move the focus of the investiga- psyche—sometimes subconsciously—makes it very difficult for
tion to solution development. Likewise, the engineer may fail to engineers that design on a regular basis to actually embrace the
adequately collect and interpret physical evidence and instead rely new set of attitudes, approaches, and processes necessary to inves-
on simplifying assumptions. These factors typically combine to tigate causation satisfactorily.
frustrate the investigating engineer and increase the probability that Given the importance of design in engineering, this is not a sur-
the failure’s cause may be identified incorrectly, potentially leading prising issue. In Beyond Failure, Delatte (2008) describes the role
to repeat failures, inappropriate rehabilitation strategies, legal chal- of engineers succinctly as “engineers design.” Further, this design
lenges, and/or skewed dispute outcomes. mentality is not limited to practicing design engineers, but runs
through the profession as a whole. Although many engineers in-
volved in construction and maintenance may not engage in design
Forensic Process on a regular basis, they typically approach problems using a design
process. Indeed, the engineer as designer is so fundamental to the
The key to determining structural causation is the application of the engineering profession that engineers rarely think of themselves as
forensic process, which aims to objectively identify the technical designers, but rather as engineers with specific design experience
cause or causes of failure by using available evidence. Essentially, and technical competency. In essence, the design mind-set under-
it is the application of the scientific method to failure investigation. pins what it is to be an engineer, is the basis of university training,
Noon (2000), in his text Forensic Engineering Investigation, states and cements the role of engineers as problem solvers.
that “a forensic engineer relies mostly upon the actual physical evi- Therefore, for the engineer wishing to pursue a career in foren-
dence found at the scene, verifiable facts related to the matter, and sic engineering, recognizing and overcoming the embedded nature
well-proven scientific principles. The forensic engineer then ap- of design is a key consideration.9 With this objective in mind, it is
plies accepted scientific methodologies and principles to interpret first helpful to consider some of the aspects of the design process
the physical evidence and facts.” that are diametrically opposed to the forensic process.
The forensic process of collecting evidence, developing failure 1. Problem solving and causation: The problem-solving aspect is
hypotheses, testing each hypothesis against the collected evidence, central to design, but distracts from the objective of forensic
and determining the most likely cause of failure, is a process of investigation.
analysis, rather than synthesis. The application of the forensic pro- 2. Evidence and assumption: Design requires putting trust in as-
cess is described by Noon (2000): “First, careful and detailed ob- sumptions that have served the profession well, but forensics
servations are made. Then, based upon the observations, a working requires that each of these assumptions be rigorously tested
hypothesis is formulated to explain the observations. Experiments against verifiable evidence for the structural failure in question.
or additional observations are then made to test the predictive abil- There is a real risk that the engineer who utilizes the design
ity of the working hypothesis.” Noon (2000) then goes on to say process on a regular basis may be ill equipped to collect and
that, “as more observations are collected and studied, it may be analyze this evidence, and/or may connect some parts of the
ceived idea of how the structure will perform. Although this have and why they fail—that separates the experienced inves-
approach is appropriate for design, the forensic process re- tigator from the rest.”
quires the investigator approach the investigation with an open 4. Keep an open mind: This is a key quality for the forensic
mind by considering many failure hypotheses, and by collect- engineer. It is critical not to reach a strong conclusion on
ing and analyzing all evidence as objectively as possible to causation early on in the investigation. This task is further com-
determine the correct failure hypothesis. Approaching the in- plicated by the iterative nature of investigations, namely, the
vestigation with a preconceived idea of the cause of failure can investigator will see evidence, collect it, develop a number
curtail the development and examination of a sufficient number of failure hypotheses, and look for evidence associated with
of failure theories and lead to confirmation bias, i.e., the inves- these hypotheses. As stated in the Guidelines for Forensic
tigator only tends to notice evidence that supports a certain Engineering Practice (Lewis 2003), “Do not approach data
hypothesis (Noon 2009). gathering with an eye toward ‘how can I find evidence to sup-
5. Synthesis and analysis: Design is a process of synthesis, and port my sense that…?’ but rather ‘how can I find all of the
there is generally no unique correct design—a number of al- important evidence?”’ Rather than asking “What caused this
ternatives may be (almost) equally satisfactory. However, in failure?” ask “What do I see?”
structural failure, there is only one correct sequence of events 5. The site visit: Evidence and familiarity with the scene of the
that caused the failure, and this sequence of events cannot be failure is critical and should not be rushed. The quality and
determined by a process of synthesis; it requires a forensic pro- completeness of evidence preservation, recording, and collec-
cess of analysis. tion will play a critical role in the strength of an investigator’s
6. Ruling out and ruling in: As discussed previously, the design opinions.
process generally commences with the design engineer devel- 6. Theoretical analysis: Beware of excessive theoretical analysis.
oping a range of reasonable concepts for a design solution. It is not unusual for the engineer to wish “to understand the
The process continues with unsuitable concepts being dropped structure.” Typically, this will be undertaken by constructing
in favor of more satisfactory concepts, to finally arrive at a sin- an analytical model or attempting to determine whether the
gle design solution. In the same way, during an investigation, structure complies with the current design code. However, both
the designer may be likely to discard hypotheses that are approaches rely on assumptions, and in failure investigation, it
considered less probable than others. However, for an investi- is the examination of these very assumptions that is critical.
gation to be considered forensically sound, this approach is Attempting to “understand the structure” in this manner can
inappropriate—the investigator must rule out each of these distract from the examination of the assumptions. As Cuoco
hypotheses on the basis of evidence. In other words, failing and Panariello (2010) point out, “Failure analysis is not the
to rule out potential causes of failure on the basis of evidence same as analysis for the purpose of design.” “Failure analysis
is just as flawed as ruling in a cause of failure based solely on must use the actual loads imposed on the structure and the real-
assumption. world capacity of the structural elements.”
7. Nature of design process: The design process is heuristic in 7. Nature of the basis of opinions: It is critical for the engineer
nature.11 In other words, the engineer is not required to scien- to probe and understand the nature of the information relied
tifically prove the validity of the design, but is required to show on to formulate opinions. Is it evidence that can be verified?
it is consistent with rules that produce structures that have gen- Is it an assumption that may be challenged as incorrect for the
erally worked in the past. Essentially, the designer asserts the structure in question? Or is it a generally accepted engineering
design is appropriate because it complies with a design code. principle that is unlikely to be challenged? This will assist
In forensics, however, simply asserting an opinion is insuffi- the engineer in assessing the strength of the opinion, and can
cient because the investigator is required to demonstrate how assist in unearthing implicit assumptions. Equally, the engineer
the opinion relies on specific evidence, scientific principles, should make sure to have a clear understanding of the infor-
and scientific logic. mation relied on to deem all other failure hypotheses less
For structural engineers with a design background, there are a probable.
number of important and useful steps that can be taken to improve 8. Noon’s “what makes a good hypothesis:” In testing the various
forensic expertise. failure hypotheses, Noon (2009) provides a helpful summary
1. Awareness of the design process: An awareness and under- of the attributes of a good working hypothesis: “All the data
standing of the design process and how it is ill suited to de- upon which it is based needs to be factually verifiable. It must
termining causation is a key first step when approaching be consistent with all the relevant verifiable data, not just
forensic engineering. selected data. The scientific principles upon which the hypoth-
2. The importance of the forensic process: Because of the impli- esis relies must be verifiable and repeatable. The hypothesis
cit nature of many aspects of the design process, constant should provide some predictive value. The hypothesis must
vigilance is required to not slip from a true forensic process be subjected to and withstand genuine falsification efforts.”12
the output of conceptual design” (Jansson and Smith 1991). With this
The importance of the forensic process, or detective skills, in de- in mind, it is interesting to pose the question: Does the adherence to
termining the cause of structural failure cannot be overstressed. simplifying design assumptions and the design process result in a form
Successful forensic investigation requires investigating engineers of fixation in the engineer when applied to forensics? Further, is the
to put aside the traditional design process and apply a forensic engineer aware that this design fixation is taking place? To date, no
process. Experience, however, suggests that such a transition is dif- experiments examining design fixation on forensic engineers have been
ficult because of the embedded nature of the design process in the conducted, but there have been numerous experiments undertaken to
engineering profession’s psyche. examine the affect of design fixation on designers, some of them en-
For the engineer interested in practicing in forensics, a critical gineering professionals, such as Jansson and Smith (1991), Linsey et al.
step is to understand that experience of the design process can be a (2010), Smith (1995, 2003), and Wiley (1998). Studies have concluded
significant impediment to actually embracing and implementing that design fixation can be empirically verified in professional design
the forensic process. Only then can steps be taken to overcome this engineers, that most people are unaware fixation is occurring, and, when
impediment. they are made aware of it, it is still remarkably difficult to overcome.
Given that this issue exists within design itself, it is a reasonable hy-
Endnotes pothesis that the same fixation occurs when the designer attempts to
investigate failure; for example, will such designers overcome design
1
Carper (2000) also highlights some of the practical differences between the fixation to challenge implicit assumptions, evaluate evidence objec-
designer and forensic engineer: “The designer is more familiar with co- tively, and put aside the traditional design process and apply a forensic
des and standards intended to prevent failure, and with the process of
process? The writer’s experience suggests that the hypothesis merits
generating alternative concepts that respond to given constraints.” “The
further investigation.
forensic engineer, however, approaches the investigation of a failure
from the perspective of physical causation and the given object, as de-
10
The term a priori is the “belief that the underlying causes for observed
signed and constructed. These two perspectives are very different and effects are already known, or at least can be deduced from fundamental
require different skills and attitudes.” principles” (Noon 2009). The forensic process is based on the scientific
2
Although the process utilizes simplifying assumptions, it is far from sim- method, and “the modern scientific method does not accept the a priori
plistic, requiring careful and meticulous consideration of a range of re- method of inquiry. The modern scientific method allows evidence to
quirements and constraints. This process is efficient, well respected, and drive theories, rather than allowing theories to drive evidence” (Noon
has served the engineering profession well. 2009). In essence, forensically sound investigations require the engineer
3 to investigate with an open mind and allow the evidence to lead the
It is little wonder that James Amrhein, quoted by Carper (2000), describes
structural engineering as “the art and science of molding Materials investigation. Noon (2009) further elaborates on this issue as the differ-
we do not fully understand; into Shapes we cannot precisely analyze; ence between empiricists and rationalists: “In short, empiricists accept
to resist Forces we cannot accurately predict; all in such a way that what is observed at face value and try to find order to explain it. Ratio-
the society at large is given no reason to suspect the extent of our nalists, on the other hand, propose up front what the order should be
ignorance.” and then sort their observations and facts to fit within that framework.”
4
At this point, a number of engineers may protest that they have success- The design process is more like the process undertaken by rationalists,
fully investigated failures in the past without the use of the forensic with design assumptions driving the investigation, whereas the forensic
process. In some situations, particularly those in which causation is process is more like the process utilized by empiricists, in which
straightforward and obvious, engineers do appear to identify causation observations and evidence are judged on their own merits. Further, with
successfully. But the process utilized typically involves the engineer respect to empiricism, Noon (2009) notes that “Fundamentally, empiri-
assuming the cause of failure and preparing a design solution to address cism is the tenet that knowledge is derived from measurements,
it—essentially demonstrating causation because the solution resolved verifiable facts, observations, and actual, as opposed to spiritual or
the issue. This is a trial-and-error approach to remediation, rather than
supernatural, experiences, feelings, or beliefs. Furthermore, as they are
a forensically sound attempt to establish causation.
collected, these measurements, verifiable facts, observations, and expe-
5
This issue is illustrated by Carper (2000) when he states that “The design-
riences are to be reported and documented without being modified or
orientated engineer may approach the investigation of a failure by sug-
interpreted by preconceived notions.”
gesting an alternative design solution.”
6
This method of identifying the cause of failure appears to be based on the
11
“The engineering method is the use of heuristics to cause the best change
premise that the design approach is the obvious cause when it is differ- in a poorly understood situation within the available resources,” from
ent than an approach the engineer has used successfully in the past. Discussion of the Method: Conducting the Engineer’s Approach to
Although such an approach may provide a clue to the cause of failure, Problem Solving, by Koen (2003). Koen (2003) also notes that “a heu-
it is not forensically sound. ristic is anything that provides a plausible aid or direction in the solution
7 of a problem but is in the final analysis unjustified, incapable of justi-
Cuoco and Panariello (2010) highlight just how different design assump-
tions can be from the performance of structures in practice: “Failure fication, and potentially fallible.”
analysis must use the actual loads imposed on the structure and the
12
“Verification of a hypothesis occurs when observations, or experiments
real-world capacity of the structural elements. Neither of these is par- and data that could show the hypothesis to be false fail to do so. This is
ticularly well represented by the code-prescribed values. Actual loads the principle of falsification” (Noon 2009).