Accident Investigation

CHAPTER6

The Analytical Process

A accident investigation is the process o f breaking down information into pieces until the investigator understands what happened; then he or she can analyze the pieces to determine ways to prevent the accident from recurring. Asking "why" is a crucial first step in discovering the causal factors of an accident.

Ca usal Analysis

Once the evidence from an accide nt is gathered, you must discover the accident sequence, and once you know the sequence, causal analysis-the process of determining the causal factors-can begin. (Senecal and Burke 1994). The goal of causal analysis is to find all of the causes including the systemic causes-not just the immediate or superficial causes. If only the superficial causes are found and dealt with, the same accide nt could happen again with a different employee.

One of the problems accident investigators sometimes have is knowing :hen to stop searching for causes. Many accident investigators have used the 5

Whys" technique to find causes. This is simply asking "Why?" five times m~mth · e root of a problem. For example, suppose John was working

59

60

Part II: Organivng 1/x lmrstigt1llo11

OUldoors carrying pipes from o n e location to another w hen . foo r. H e re 1re five "\'vh y?" questions the investiga.10 ,: pipe fell on his going deepe r imo Lhe cause of the accident: r cou ask John, each

Investigato r: J o hn, why did the pipe fall o n your foot?

John: I dropped it.

lm·estigaror. \Vhr did yo u drop it?

J o hn: It slipped out of my hand.

Investigaror. \\7hy did it slip?

J o hn: It w as wee.

Investigator. \'vhy was it wet?

J o hn: The pipes were sitting in a pool of water.

ln vestigator. \'lhy were the y sitting in wate r ?

John: It rained earlier today.

(Obviously, th e questioning process in this case has o nly begu n.)

Causal analys is is a process in which an in vestiga tor an alyzes, probts, di scovers, ponders, and uses scenarios, facts, tests, and assumptions to determine what caused an accide nt. Cau ses and cau sal factors can exist at many levels-worker, equipment (failure or hazard) , supervisor, management, management sys tems (policies and procedures) , and even corporate culture, philosoph y, and style. Lower -lev el causes are usually more specific ro one particular accident, buc th ey are still important to list and fix . In fact, p~blt111S ar th e lowest lev el can lead to worthwhil e engineering and eqmpmenr co rrections that solve a problem o r elimina te a h azar d. Upper-level causal fac to rs are more difficuJt to fix, but doing so will affect a broader range of people and situatio n s and help to prevent future accidents (DO E 1999), Exhibit 6.1 displays the level s o f accountability for accide nt investigations.

Causal Analysis Example . ef An accident occurred JJJhrn a worker did 110/ II.ff a fork and tag to isolate a pure equ;p11m1I and k eep the electriri!J out.

Chapter 6: The A11a!J•timl Proms

Exhi bit 6.1

worker or 1. equipment level

~isorlevel

~anagement level

4. Corporate level

LEVELS OF ACCOUNTABILITY

This is the lowes t level of accoun_tabiHty. At the work er level work is p erform ed and eqwpme~t operat~s. Causal factors in this area include equipment failure s, · adequate training, inexperience, and what many ~:nsider human error (which would include training, experi ence, etc.).

People at this level describe how work is to be done. Causal factors often include inadequate handling of job safety analyses, com~~nication, or scheduling, and lack of proper superv1s1on .

Management level d ictates policies and proce- dures. Cau sa l factors at this level are usually related to budget issues, communication, and policies/ procedures.

This highest level dictates the culture, ph ilosophy, and style of the company. If problem s are found and corrected at this level, many accidents ca n be prevented.

A lower-level causal factor is that the worker failed to use the proper lockout/ ugout procedure to isolate the energy. A hig her- level facror could be that lllln agement did not enforce or ha ve a policy o n lockout/ tagour. If the lower- b-el problem is fixed- the worker is Lrained on lockou t/ tagour procedures and given a lock and tag-that worker probably will not have another accident. However, if management develops a policy of training al/workers in lockout / tJgout procedures, man y similar accidents can be prevented .

With this type of accident, a causal factor may also exist at the corporate lerel if upper manageme nt fail ed to audit the plant 's procedure s and th

erefo re did not find out that th e plant lacked loc kout / tagout policies. If upper man age ment started performing p o licy audits, many m o re rypes o f accidents cau sed by failure to follow polky or procedure-not ju s t electrical and lockout/ tago ur accidenrs--could b e avoided.

61

Part II: Or:._.1,11111 ::;p,g tbt l m"tshga/1011

62

Thi s example demonstra1cs 1.he imponancc of I . all ferris. If yo u investig:uc only a1 the lowe r levels y~:1~~z~ng_ an accident it cau sal facto rs-and thu s t.h c corrective actions-at th hi \ 1111.ss sorne of the changes at higher le vels is more com p lex tha kine g 1cr lc\'els. Ma king It is difficult to change a compan),'s c ul ru n hil~a g them at lowt r ln•ds·

re, P osophy, or style . no t mean that yo u should avo id inves ti gaung high-I , ·I . llus d0c$ I , I f . . . e\ c causal fact e , c o accountability 1s important-b ut everyone sho Id b ors-each

o nl y corrective action th at is effecrive is o ne that is .. u. de aware that the b y the appropriate decisio n-makers. inmate and supponcd

Hazards vs. Failures

As yo ~ Lm ·esrigate ac~ide nrs, it _is im po rta nt t0 distinguis h between hazards and fail ures. A hazard 1s something th at has the potential to cause injury, and haza rds are correctable. Examples of ha zard s are a sharp table edge or a pool o f grease o n the tloor.

r\ f aihm is something th at goes wro ng with perso nnel, equipment, or the en vironment (Ferry 1981 ). A failu re ma}' or ma y no r have th e potential 10 cause in jury. If it does, it is a1 so considered a haza rd. A dead banery in an automo bile is an example of a failu re tha t is no t a haza rd ; ordinarily it would no t cause an accide nt. H owever, an au tomobile tire tha t fails and blows out while someone is driving is consid ered a hazard as we ll as a fail ure, bec:iust the failure could cau se an accident.

F::Lilures are usuall y caused by fau lty design, a defect, inadequate maintcruncc, limits that "vcre exceeded, or envi ronmental effects.

Analytical Techn iqu es

i\bn }' accident investigation tec hniques were de rived fr om sys tem s 3feiy

tech niques that we re developed to an alyze eq uip m en1 design failu res and hazards. These types of techniques are used by the D cpartmc n1 of Defense

and D epartm ent of Energy (Vincoli 1994).

Chaptrr 6: Thr A 11,1b'timl Proms

. 'd t investigation tec hniques. They are in troduced . five 01:un acc1 en . fhcrc arc.: . din depth in the next part of this book:

here :i nd w1U be exam ine . d causal facwrs analys is

, e\'cnts an , ch:inge anal ysis

, bafl'lc r analysis

' :in:tlrucal uc es • c:1usc an d effect ana lysis

Each technique analyzes a differe nt type of problcr_n, and each has stre ngths d weaknesses. The techni q ues are broad e nough in scop e tO han dle sm all

;:cidents as well as m ajor ca tas tro ph es. Usi ng. s~vc ra l tech~iqucs_ in a n

111 \·estiga tion ensures accuracy, consistency, ~nd validity and helps 1m·es agato rs

co obtain more info nnatio n abo u~ the acc1de ~t .s_e~ue n~e, be m o re ac.curatc ,.nd precise, an d share inves tigaave res~o n s1biliaes_ w1t.h o th e rs. It 1s also

ssible for the res ults of differe n t techruq ucs tO va li date each o th er. 1l1ese :hniques must not be used mec hanically o r wi th ou t con sideration of the awdent sequence and circumstances (DOE 1999).

A flowchart of analytical tech niques is ill usmned in E xhibit 6.2.

Benefits of Using Analytical Tech niq u es

• If you do no t use analytical tec hniq ues, it is ve ry easy to fi nd only lower- le\·el causal factors an<l ,niss the syste mic factor s.

• Usingana.lrtical techniques for every accident investigation lends consiste ncy to yo ur safety program.

' Analytical tec hnic1ues will he lp yo u to ma ke a s moo th a nd co n siste n t transition from facts to causal factors .

' The thoroughness of the analytical tec h niques wi ll give vou confidence iba t ro ur invesugatio n dete rm ined wh at rc:-tll y happe ned a nd that your rtrommcndations wi ll prevent future acc id ent s.

63

64

Port II: O,g,,uuZf,,g the lnrtshgaho11

Exhibit 6.2

ANALYTICAL TECHNIQUES FLOWCHART

Has th ere been a chang e in the process?

Is the task recurring? Has it been don e

accident -free?

Change Analysis

ls th ere a need to evaluate the loss control efforts or

~::g~;ef Time Loss Analysis

Is there a failure of a system or hardware? Could a systema tic

evaluation of failure modes be usefu l?

Failure Modes and Effects Analysis

Has there bee n a deviation from the

requi reme nts? Can a review of the codes, requiremen t s, and

standard s be us eful ?

Design Criteria Analysis

Is there a need for a time-ba sed matrix of all personnel at the

accide nt scene?

Is there sti ll a void In the information? Can a

speci fic test or tec hn ique fi ll th is void?

Other Speclalind Techniques

Ad apted from OSHA In stitute 1995

Chapter 6: The A mi(yiral Prows

Summary

or circumstances that contribute to an accident are called causal flu.' eren:rreccion of these causa l factors_ at whatever I,evcl th~y may oc~ ur factors . C rcicular incid ent will not recur. \Vhe n u sing analyacal is whai ensur~s th ac_ a p,•ors must b e able ro disti nguis h betwee n ha zards

. es mves uga _ _ . cechru~u ' U . causal analysis and anal ytical tec hruques will ma ke the lf!Jf.ulure s. sing _ 1n\·esug1ti on more effecuve.

REVIEW QUESTIONS

I. What is a ca usal factor?

2 Describe the four leve ls of accountability.

3. What is the difference between a hazard and a fa ilure?

4. Why is it advisa ble to use more th an one analytical tec hn ique to investigate

an acc ident?

65