Semi-quantitative risk analysis for underground development projects

P. N. Standish Safe Production Solutions P. A. Reardon Safe Production Solutions ABSTRACT: The application of semi-quantitative Risk Management techniques to large underground developmen t projects involving over 10kms of tunnelling is described. This approach involved a top level risk assessment which led to specific detailed Risk Analysis. The results of these included ranked hazards and implemented management plans. A qualitative broad brush technique was used to find potential hazards that: Were not identified by design or operations personnel using traditional desk top risk assessment techniques; Could be ranked for importance using a team approach; and Assisted in focussing further design and contingency Risk Management approaches. Specific Risk Management activities are also described where a combination of qualitative and quantitative tools were utilised, including: Team based risk assessment; Fault tree analysis; Event tree analysis; and Control development. Use of semi quantitative tools allowed for a comparison of the importance of hazards: Between different job sites; At different times during the life of the project; and Against government / society standards of acceptability.

1 INTRODUCTION The assessment of risk in tunnelling projects can allow project management teams to demonstrate the rigour of their designs and plans. An approach that provides a dual focus involves taking a systematic approach with a team of people who are objective, trained and experienced in the subject area. A team works through the issues around the project and takes a helicopter view of the site. This process is aimed to identify potential catastrophic losses (those that can potentially stop the project through their action on the People, Environment, Assets or Reputation). This broad focus analysis is then followed up with more targeted risk assessments on specific subject areas.

Both levels of risk assessment are conducted in a manner that is systematic and aligned with various loss models. 2 STAGES OF THE RISK ASSESSMENT The key steps in the conduct of a risk assessment include: Defining the scope and goals; Collecting sufficient background information; Forming an appropriate team t o provide technical information and allow for an objective assessment; Utilising a facilitator and suitable team methods; and Documenting the results.

3 BROAD FOCUS RISK ASSESSMENT This broad focus level of risk assessment allows for identification of the show stoppers in a project. Ideally the best time to conduct this level of analysis is during the pre-feasibility stage. Doing an analysis at this time can increase the level of confidence in approving groups. Figure 1, after AS4360 Risk Management, shows the way in which the process is applied and how it fits within a continuously improving framework of Plan Do Check Act (Hutchison, 1997).
Establish the context

approach is in the identification of additional hazards through the use of the alternate processes. Historical information is gathered through reference to site information, published records and g o v e r n m e n t a n d S a f e P r o d u c t i o n S olutions databases; Brain storming processes used are based on creative thinking strategies (de Bono, 1985) and association with key words that are intended to trigger thoughts on additional loss scenarios by the team. Flow chart development uses a chronological development model to identify the steps in the project. This can also be based on the PERT charts for the project schedule but is typically developed as shown in Figure 2. PERT charts are typically too detailed for the broad focus risk assessment and a higher level consideration of the project better suited to the development of catastrophic losses. Steps identified are considered again in the loss matrices and Input Process Output tools.
A - DESIGN

Communicate and consult

Identify risks

Analyse risks Evaluate risks

Assess risks

Treat risks

Figure 1 Risk assessment Framework (a fter AS4360)

A1 - Conceptual Design (Based on Lift 1) A4 - Issue for A2 - Detailed Design Construction A5 - Data Acquisition, Includes Financials and Geotechnical Information

3.1 Broad focus identify risks The types of loss that have been identified by the authors and the project teams with whom they have been working have included: Major ground fall events; Inrush of water or flowing strata / ground; Fire(s); Traffic losses; Impact on external infrastructure or groups; Emergency Response failures; and Clashing effects. Methods used in achieving this identification are based on: Historic records of similar tunnelling methods; Brain storming techniques; Flow charting (Joy 1994); Loss matrices; Input Process Output models (after Joy 1994); and Fault and Event Tree Analyses (after AS3931). The logic of this approach is to provide redundancy in the analysis. Many similar losses are identified in each process and the value from taking the

Monitor and review

A3 - Design Approvals

B - PREPARATION FOR SITE

B1 - Establish management teams

B2 - Bid and Award Contract

B3 - Preliminary u/g works and provision of services (Pybar)

B1(a) Recruitment

B4 - Initial surface works.

Figure 2 Typical Flow Chart

Loss matrices involve a mechanism for developing credible scenarios by considering each type of catastrophic loss identified during the research phase and mapping this against the steps identified by the team through the flow chart tool. Xs in Figure 6 indicate the types of credible loss to be considered in subsequent analysis.

Establish Mgt Team -

Bid and Award Contract -

Preliminary U/G Works X

Data Acquisition X

Confirming that underlying losses are appropriately considered in the ranking process. Broad focus development of controls

Inrush

3.3
X X

Major Collapse

Conceptually, a control can be considered an intervention that aims to decrease the probability of an occurrence, and/or minimise the consequence of the loss event Control of hazards involves reference to two models - the Nertney Wheel and the interventions system model.

Etc.

Figure 3 Typical Loss Chart

Input Process Output models may also be used to assist in determining the types of losses that could occur on the project. Figure 4 shows this approach graphically. Inputs and outputs are described, the flow chart steps are considered then potential losses are generated for each of these.

The "Nertney Wheel" model illustrates how People, Equipment and Work Methods react to the Working Environment. From this standpoint control can only be accomplished by addressing all the related factors on a case-by-case basis. The control action model (Joy 1992), on the other hand, illustrates that controls may be preventative, monitoring, provide a first response or help the recovery from a loss.
Workplace

INPUTS

OUTPUTS

PROCESS Hardware People

Safe Production

Figure 4 System Model

INPUTS

OUTPUTS

Fault and Event trees may also be used. Dis cussion of these is not presented here and their use in providing a quantitative assessment (where data is available) is still a developing science for tunnelling applications. The use of higher level generic fault trees is given later in this paper. 3.2 Broad focus analyse risks At potentially catastrophic levels of loss the philosophy of treating unknowns as catastrophic until proven not so is made. This approach is aligned with that of the Safety Case approach m a n d a t e d f o r p e t r o c h e m i c a l a p p l i c a t i o ns (Department of Industry Science and Resources, 2000). Analysis of the losses is achieved through a process of: Developing a generic fault tree; Using a semi-quantitative matrix to rank the losses; and

Methods

Environment

Waste Losses Injuries

Figure 5 Wheel of Safe Production

Equipment manufacturers and operations that use the equipment can determine and install interventions that aim to be effective in minimising the chance or consequence of a loss event within this system. Combining the principles of intervention with Nertneys wheel gives us a framework or a control matrix, to categorise existing or proposed controls: Controls also have varying level of effectiveness, which in the first instance are the so called "hard" and "soft" controls as mentioned in a range of statute.

4 SPECIFIC RISK ASSESSMENTS This level of risk assessment "homes in" onto specific risks that have a potential to do serious damage in a project and is conducted after the pre feasibility stage. In many cases, tools such as FMEA, Fault Tree Analysis, Human Error Analysis and the like, depending on the actual operation, are used. FMEA or Failure Modes and Effects Analysis tool was developed by NASA to assist in identification of potential problems in rockets etc, and provides a method for conducting a rigorous review of a machine system. The method uses a table developed by considering each item of equipment and examining how it can fail. Where a failure mode is controlled by a protection system - that too is analysed to determine impacts of failure. Fault Tree Analysis is a systematic tool that leads to the logical development of the many contributing failures to one unwanted event. All the potential scenarios that could lead to this event are listed as the first tier of branches below the unwanted event. Similarly, the contributing events leading to each of the potential scenarios are identified and placed in the next tier. This process is repeated until all the generic groups of loss are identified. Human Error Analysis includes consideration of skill, knowledge, application of rules, slips, lapses, operating when not fit for work, misreading dials, etc. As may be expected, the contribution of human error to the risk depends on the extent of the automation available and is generally lower for well automated operations.
A B C D E

Table 1 Probability Ranking

Category Definition Common or frequent occurrence (1:yr) is known to occur or it has happened (1:10yr) could occur or have heard of it happening (1:100yr) not likely to occur (1:1,000yr) practically impossible (<1:10,000yr)

Table 2 Consequence Ranking

People 1 2
Fatality Disabled

Equip ment
$10m $1 10m

Product ion
2 wks 1d 2wk 1 shift 2 hrs <1hr

Environment
Major Serious Moderate Minor Insignificant

3 (weeks) 4 5
LTI Minor

major LTI $100K $10k $1k

Table 3 Risk Ranking Combined Consequence and Probability

Probability A B 2 5 9 14 19 C 4 8 13 18 22 D 7 12 17 21 24 E 11 16 20 23 25

Consequence

1 2 3 4 5

1 3 6 10 15

4.1 Specific risk assessment controls The assessment of specific risks is done quantitatively and is defined as the combination of the probability of the event and the maximum reasonable consequence of the event.

Risk Level

High

Med

Low

From these assessments, the specific risks now ranked semi-quantitatively are evaluated by the team and appropriate actions instituted to ensure prevention of the event or events. These controls are typically addressed through consideration of the key elements of the Wheel of Safe Production (Figure 5) and the nature of a controls intervention action Prevention, Monitoring, 1st Response and Recovery.

This leads to a control matrix , an example of which is given in Figure 6 (where (H) and (S) refer to hard and soft controls respectively).
Prevent Weathered zone shotcreted (H) Perimeter security fence (S) Security lighting (S) Evas on surface (H) Design and fit O/H protection for kibble (S) Monitor Security patrols. (S) 1st Response / Recovery Emergency Response plan (S)

Hutchison, Douglas, Dr. Safety, Health and Environmental Quality Systems Management. Lanchester Press Inc, Sunnyvale, USA. ISBN 1-57321-008-0 Library of Congress No 96-79260. 1997. Joy, Jim CCH / ALARA Workplace Risk Assessment and Control Manual. CCH Australia Ltd ISBN 1862 64 565 5. 1994. NSW Department of Mineral Resources Risk Management Handbook for the Mining Industry - MDG1010, May 1997 NSW Department of Mineral Resources Guide to Reviewing a Risk Assessment of Mine Equipment and Operations MDG 1014, July 1997 Standards Australia AS3931 Technological Risk Management Systems. 1998. Standards Australia Management, 1999 AS4360, Risk

CWE1

Regular checks i . e . c or r os i on (S) Communicati on reliability (S) Cage lighting (S) Emergency Response plan (S)

Equipment

Standards for hand tools ((S)) AS1457 referenced (H)

R e gul a r s ha f t inspections (S) Independent security checks (S) Consider wheel on end of stabiliser for shaft inspections (H). Winder driver trained to remove un authorised people (S)

Procedures
Trained in identifying potential falling object hazards. (S)

People

Emergency Response plan (S)

Figure 6 Example of a Loss Control Matrix

5 CONCLUSIONS Semi-quantitative risk assessment provides a mechanism for project management teams to: Increase their level of understanding of potential losses; Demonstrate the rigour of their considerations to venture capitalists, regulators and other stake holders; Get on the front foot in the development of controls to minimise project cost; and Manage the left field losses that frequently are faced in tunnelling operations. 6 REFERENCES De Bono, Edward Tactics, The Art and Science of Success. Fontana Harper Collins Publishers. 1991

CWE Controlled Work Environment