Project risk management processes techniques in sights phần 5 ppsx

However, the negative dependence introduced into the activity duration relation- ships by contingency planning induces strong, positive dependence between associated costs. If A costs more than expected, B tends to cost very much more than expected, because of the need to keep the project on target, quite apart from other market-driven sources of dependence. Put another way, cost and duration modelling of uncertainty that does not explicitly consider contin- gency planning tends to estimate time uncertainty erroneously (usually optimis- tically) and fails to structure or explain it and tends grossly to underestimate direct cost uncertainty. Considering the impact of contingency planning will clarify apparent time uncertainty and increase apparent direct cost uncertainty. Common causes of knock-on effects are design changes and delays, which not only have a direct impact but also cause ripple effects termed ‘delay and disruption’. Often direct consequences can be assessed fairly readily in terms such as the number of man-hours required to make a change in design drawings and the m an-hours needed to implement the immediate change in the project works. Ripple effects are more difficult to assess and may involve ‘snowballing’ effects such as altered work sequences, conflicting facility and manpower requirements, skill dilution, undetected work errors, and so on. Example 8.4 Widening fire doors causes substantial delays In 1991 apparently small cha nges in the design of fire doors on Channel Tunnel rolling stock was expected to lead to a delay of up to six months in providing a full service for car and coach passengers, substantially reducing expected revenue for Eurotunnel, operators of the tunnel. The problem was caused by the insistence of British and French authorities that the width of the fire doors separating the double-deck car shuttles should be widened from 28 to 32 inches (Taylor, 1991). Example 8.5 A delay-and-disruption claim Cooper (1980) has described how a computer simulation based on influ- ence diagrams was used to resolve a $500 million shipbuilder claim against the US Navy. By using the simulation to diagnose the causes of cost and schedule overruns on two multibillion dollar shipbuilding programmes, Ingalls Shipbuilding (a division of Litton Industries Inc.) quantified the costs of disruption stemming from US Navy-responsible delays and design changes. In the settlement reached in June 1978, Ingalls received a net increase in income from the US Navy of $447 million. It was the first time the US Navy had given such a substantial conside ration to a delay- and-disruption-claim. 146 Structure the issues The need to appreciate fully the implications of knock-on effects in a project is clear, especially for activities late in an overall project sequence that may be considerably delayed, with possible contractual implicatio ns of great importance. As Example 8.5 illustrates, this process of apprecia tion can be greatly facilitated by appropriate diagramming of activity–source–response structures and their interdependencies. Develop diagrams The use of a range of diagrams is advantageous throughout the structure phase to document and help develop insights in the struct uring process. Precedence networks and Gantt charts are key documents because they capture key aspects of the project base plan. However, other diagrams are important in term s of capturing a range of wider considerations. For example, if a formal model is used to link Gantt charts to resource usage and associated resource constraints, these issues will require appropriate diagrams. If direct/indirect cost models are used, other stand ard diagrams will be required. Of particular concern here is diagrams that summarize our understanding of source–response structures, and links between activities, sources, and responses. Ensuring that the earlier steps in the structure phase result in a set of diagrams that summarize the classification, ordering issues, and then linking them is ex- tremely important. Complexity is inherent in most projects, but it must be made manageable to deal with it effectively. A summary diagram structure, which all those who need to be involved can discuss as a basis for shared understanding, is very important. Organizations that have used such diagrams often stop doing so because they are difficult to const ruct, but start using them again because they realize these diagrams are difficult to produce precisely because they force a proper disciplined understanding, which is otherwise not achieved. One such diagram is the source–response diagram of Figure 8.2, which was initially devel- oped for offshore oil projects and subseque ntly adopted by a range of organizations. In principle, a numbering system of the kind described early in Chapter 7 (u, v, w, x, y, z designations) could be used to drive a computer-generated version of Figure 8.2. However, manual approaches, with computer graphics when appropriate, have been employed to date. Example 8.6 Source–response diagrams for an offshore platform jacket Figure 8.2 provides an illustration of source–response diagrams in the context of the fabrication of an offshore project jacket (the structu re that Develop diagrams 147 Figure 8.2—Source–response diagram From Chapman et al. (1987), reproduced by permission of John Wiley & Sons ). sits in the water to hold production and accommodation facilities)—this is the first section of a diagram that continues in the same vein for several pages. The ‘7’ in the large triangle indicates this is the start of the diagram for activity 7 (jacket fabrication). The ‘7b’ label at the end of the diagram’s horizontal centre line indicates a continuation to a further page (diagram section), which will start with ‘7b’ on the left-hand side. Primary sources are represented by circles along the diagram’s horizontal centre line and linked parallel lines. The first source (labelled 1) in a time- of-realization sense is ‘yard not available’, because another jacket is still under construction in the contracted yard (a dry dock construction area like a big shipyar d), and our jacket has to await its completion. A close second in this time-of-realization sense (labelled 2) is ‘mobilization problems’: we can get access to the yard, but it has not been used for some time, so it will take time to get up to speed. These two sources are mu tually exclusive: we can have one or the other, but not both—this is why they appear in parallel. All the other sources are in series, indicating they can all occur, without implying additive or multi- plicative effects at this stage. Their sequence is nominal. Dependence relationships could be indicated on the diagram and lead to ordering sources, but independence is assumed with respect to those sources shown. Links in this diagram are limited to links from earlier activities discussed in notes along the top of the diagram. Links could appear as arrows between sources and responses, with links out to other diagrams if appro- priate. Identification of all these links, dependence, and ordering issues is part of the structure phase steps identified earlier. Responses are represented by boxes, ordered to reflect the preferred implementation sequence. Secondary sources are represented by circles at the side of the primary responses. For example, if the yard is not available, the preferred response is to ‘mobilize’ (get ready to start work, making temporary use of another site) and ‘accept a short delay’. The secondary source here is a ‘longer delay’, which would lead to the second- ary response ‘find an alterna tive yard’. The secondary source here is ‘none available’, at which point ‘mobilize’ and ‘accept a longer delay’ are the only remaining option. These responses and secondary sources illustrate further the complexity of the generic types of response we may have to consider to capture the most effective response to uncertainty. They also make it clear why a diagram to capture the structure provided earlier is a very good test of understanding, which may lead to redefinitions in earlier steps. The final source on the last page of the source–response diagram for each activity is a collector/dummy risk that represents residual uncertainty after specific responses. The ordered boxes that appear below this residual Develop diagrams 149 uncertainty collector are the general responses. The importance of the structuring process as a whole is highlighted by the need for this feature. It also indicates that the residual uncertainty of real interest is the combined effect of all individual sources (net of specific responses) less the effect of general responses. This serves to emphasize further the importance of structure. Implicit in the identify phase is a very complex decision tree that will remain an implicit, ill-understood ‘bushy mess’ unless the structure phase is pursued until source–response diagrams like that of Figure 8.2 can be drawn. Completion of such diagrams by risk analysts, and subsequent verification by all relevant players on the project team, is a watershed in the overall RMP. Fault trees and event trees Two common approaches used in a system failure analysis context that underlie the Figure 8.2 approach are fault tree analysis and event tree analysis. It can be useful to adopt these approaches in their basic or standard forms as a preliminary or an alternative to the use of source–response diagram formats like Figure 8.2. A good classic reference is NUREG (1975). Event tree analysis involves identifying a sequence of events that could follow from the occurrence of particular source–response configurations and then repre- senting the possible scenarios in a tree diagram where each branch represents an alternative possibility. In fault tree analysis the process is reversed, working backward from a par- ticular event known as the top event, in an attempt to identify all possible sequences of events giving rise to the top event. Ishikawa or fish bone diagrams (Ishikawa, 1986) adopt a similar ap proach, showing necessary inputs to a particular final position. Influence diagrams In event tree and fault tree analysis there is still the problem of ensuring completeness in the set of possible failure modes included. A more versatile representation of causes and effects can be achieved with influence diagrams, as used in ‘systems dynamics’ (Forrester, 1958, 1961; Richardson and Pugh, 1981; Senge, 1990) and ‘cognitive mapping’ (Eden, 1988) . One advant age of influence diagrams over tree diagrams is that much more complex interactions can be shown, including feedback and fee dforward loop effects. 150 Structure the issues Example 8.7 Cognitive mapping shows the knock-on effects of design changes Williams et al. (1995a, b) describe the study of a large design-and- manufacturing engineering project, undertaken as pa rt of a delay-and- disruption litigation. Design changes and delays in design approval would have caused delay to the project. In order to fulfil tight time con- straints, management had to increase parallel development in the network logic, reducing delay but setting up feedback loops that markedly increased the total project spend. Cognitive mapping using specialist computer software called ‘Graphics Cope’ was used to elicit the rela tionships. The cognitive map contained some 760 concepts and 900 links. Over 90 pos- itive feedback loops were identified, illustrating the complex dynamics of the real situation. Figure 8.3 summarizes some of the key feedback loops. The situation in Example 8.7 is similar to that described in Example 8.5. It is unfortunate that the very considerable benefits of constructing cognitive maps to explore source–response dependencies were sought after these projects got into serious difficulties, rather than before. Develop diagrams 151 Figure 8.3—Key feedback loops in Example 8.7 Reproduced by permission of the Operational Research Society Influence diagrams such as Figure 8.3 are essentially a qualitative tool, although they can provide a starting point for quantitative, systems dynamics models (Rodrigues and Williams, 1998; Eden et al., 2000; Howick, 2003). They do not indicate the magnitudes or the timing of influence relationships that would be quantified in systems dynamics model simulations. Thus a link between two factors X and Y does not in dicate the strength of the link: whether it is continuous or intermittent, or whether the impact on the influenced factors is immediate or delayed. Nevertheless, an influence diagram can be a useful aid to understanding a complex situation, particularly if effectively inter- preted. It explores positive and negative feedback loops in a way Figure 8.2 does not accommodate directly, providing a very useful complementary or alternative technique. Diffenbach (1982) suggests a number of guidelines for interpreting influence diagrams: 1. Isolated fac tors. A factor not linked to any other factor suggests either that the isolated factor is not relevant to the depicted situation or that not all important links and factors have been identified. 2. Influencing-only factors. A factor that influences other factors but is not itself subject to influence from other factors prompts questions about overlooked links and factors that might influence this factor. 3. Influenced-only factors. A factor that does not influence any other factors prompts questions about overlooked links and factors by which this factor might influence. 4. Secondary and higher-order consequences. Chains of influence suggest poss- ible secondary and higher-order consequences of a change in a given factor in the chain. 5. Indirect influences of A on B. Chains can reveal potentially significant indirect influences of one factor on another. 6. Multiple influences of A on B. One factor may influence another in more than one way. These multiple influences could be direct (by link) or indirect (by chain) and of the same or opposite sign. 7. Self-regulated loops. A chain with an odd number of negative links that cyc les back to meet itself is a self-regulating, negative loop. Successive cycles of influences result in counteracting pressures. 8. Vicious circles. A chain with zero or an even num ber of negative links that cycles back to meet itself is a self-reinforcing, positive loop. Since it is unlikely that vicious circles will operate endlessly, unresisted by countervailing forces, one should look for one or more negative loops that are interrelated with the positive loop by means of a common factor. The process of construction and interpretation of influence diagrams goes beyond identification of direct source–response and cause–effect relationships. It also assists in identifying potentially important links, such as the nature of source–response chains associated with vicious circles, or particular sources 152 Structure the issues that influence many other sources either directly or indirectly. Increased under- standing of cause–effect relationships can also promp t the formulation of additional responses. More general soft systems models The use of influence diagrams can be viewed as a special (reasonably ‘hard’) version of a range of ‘soft’ approaches usually referred to as soft systems, soft operational research, or other labels that span the two, like problem structuring methods (Rosenhead, 1989; Checkland and Scholes, 1990). All these ideas are directly relevant to the structu re phase. Structure fit for the purpose? As with other phases of the SHAMPU process, the structure phase is itself an iterative process. In particular, we cannot assess the importance of some sources until we have identified responses and considered possible inter actions between sources and responses. However, some prior assessment of the importance of identified sources is necessary to guide the initial structuring, to avoid too many or too few sourc e and response categories. The structure phase clearly links to all previous phases, because it is a form of robustness analysis associated with earlier phases, as well as ordering issues for subsequent phases. In particular, changes to the structure phase outputs may be triggered by later changes to identified sources and responses. Figure 4.1 limits the feedback loops assumed to two from the evaluate phase and one from the manage phase, but the impact of the obvious linkages here in terms of selectively revising earlier structuring is important. However, the structure phase should always be as complete as possible given the progress made in the identify phase before moving on to the ownership phase. Conclusion The structure phase as described here is a very important part of the SHAMPU process. It is about transforming the information generated earlier into a quali- tative model of project uncertainty, ideally summarized in diagrams, with under- lying computer-based mod els to handle changes where appropriate and feasible. The richer the information generated in the identify phase the greater the need for care in the structure phase to provide a sound bas is for inferences to follow. In the authors’ experience some key points to bear in mind in the structure phase are: Conclusion 153 1. independence, or lack of it, is one of the most important assumptions made in any modelling of uncertainty; 2. in a cost dimension high levels of dependence are endemic, and in an activity dimension important instances of dependence are endemic; 3. making inappropriate assumptions about dependence or avoiding quantifica- tion because of dependence are potentially dangerous cop-outs that may negate the whole process—it is the difficult bits that can be particularly important; 4. the most effective way to understand uncertainty dependence is to model it in causal terms; 5. ‘statistical’ dependence is best thought of as a causal dependence of several kinds that cannot be sorted out or that it is not cost-effective to sort out at this stage; 6. ensuring a simple but effective structure for sources and responses as well as activities is greatly facilitated by diagrams like Figure 8.2; 7. being prepared to experiment with differ ent forms of diagram, like Figure 8.3, can greatly enhance the RMP as a whole. 154 Structure the issues Clarify ownership9 It is an equal failing to trust everybody and to trust nobody.—18th century English proverb Introduction In principle, making sure that every source of uncertainty and all associated responses have a manager and an owner in financial terms (possibly different parties) is recognized as basic good practice. In practice, this worthy ambition is not often achieved. One obvious reason for this is a failure to identify issues early in the Project Life Cycle (PLC) that later prove to be a serious source of difficulties or a serious lost opportunity. Another is a failure to identify relation- ships between issues that prove to be important. These are fundamental failures in other phases of the Risk Management Process (RMP), not failures of the ownership phase per se. However, even if issues are duly identified and links between them appreciated, effective management of these issues requires appro- priate and effective allocation of issues to those parties involved in a project. This is the focus of the ownership phase in the SHAMPU (Shape, Harn ess, and Manage Project Uncertainty) process. Failures of risk management associated with the allocation of ownership of issues tend to arise because this activity is not recognized explicitly, or not given sufficient attention. Issue allocation always occurs in any situation where more than one party is responsible for the execution of a project. Just as roles and responsibilities are allocated to parties concerned, so too are uncertainty manage- ment issues associated with the enterprise. However, allocation of issues, and consequently risk, can take place by default and need not be explicit, intentional, or clearly articulated. The consequences of an allocation, particula rly a default allocation, may not be fully appreciated, and the manner in whic h allocated issues are to be managed may be unclear, if they are managed at all. This chapter attempts to provide a framework for efficient and effective issue allocation processes, in terms of an explicit ownership phase in the SHAMPU process. Locating the ownership phase after the structure phase of the SHAMPU process is appropriate because it is in some respects a particular kind of structur- ing. Locating the ownership phase before the estimate phase of the SHAMPU process is appropriate because some ownership issues need attention before starting the estimate phase, although some ownership phase tasks can be com- pleted quite late in the SHAMPU process. [...]... their management in more detail Example 9.1 Risk analysis in a competitive bidding context Consider the nature of risk analysis carried out in three closely related but quite different contexts: risk analysis by the client prior to putting a contract out to competitive tender; risk analysis by a bidding contractor prior to submitting a tender; post-tender risk analysis by the winning contractor In. .. the project team to accept responsibility for physically managing this issue, including developing procedures and plans to avoid the issue being realized and developing contingency plans should it happen It might have been worth indicating to those responsible for avoiding the issue what sort of unpleasant futures might be forthcoming if the issue was realized But there was no point in making the project. .. issues is given in Curtis et al (1991) Example 9.2 Risk allocation in management contracting systems of procurement Management contracting systems involve the employment by the client of an external management organization to co-ordinate the design and construction phases of the project and to control the construction work The concept underlying management contracting systems is that the Management Contractor... £20 million in the budget In addition to distinguishing between project management and board-level financial responsibilities, some organizations are moving toward distinguishing between issues owned financially by the project manager, issues owned by those 164 Clarify ownership at the sharp end of specific aspects of the project, and issues owned by a number of intermediate management levels, in the context... designated intervals, working to one significant figure For size the uncertainty purposes it is convenient to assume that all values in a particular interval are equally likely 5 Optimistic outcome probability Step 5 involves assessing the probability of an outcome in the interval centred on the optimistic outcome scenario value In the context of a threat, given the nominal 10 percentile interpretation... these principles are duly acknowledged by the appropriate personnel Identify possible issue owners The second step in scope the contracting strategy involves identifying parties who could be expected to own some sources of project- related uncertainty and associated responses An obvious starting point is the list of key players identified in the SHAMPU define phase As noted in Chapter 5, this list includes... be usefully driven by risk management considerations as well as the physical nature of the project works For example, a large construction project that Chapman was associated with involved an RMP that included examining all the major components of the project in relation to their key sources and responses with a view to minimizing the number of issues that would require managing across contractor boundaries... the project works This is based on the proposition that foresighted, co-operative management of the interactions between the parties can shrink the risk (and risk inefficiency) inherent in construction work The NEC main options offer six different basic allocations of issues between the ‘employer’ (client) and contractor, and whatever variations in strategy between different contracts within a project. .. the estimate phase Sizing the uncertainty The next task, size the uncertainty, involves making ‘sizing’ estimates of potential implications for project performance by roughly assessing the range of possible Sizing the uncertainty 177 outcomes and associated probabilities Example 10 .5 on p 180 illustrates what is involved, but first consider the description of each step summarized in Figure 10.1a 1 Pessimistic... the scope of the risk analysis undertaken will be in uenced by the predominant concerns of the party undertaking the risk analysis and the information about uncertainty available Analysis by the client—Uncertainty is evaluated and the project design is developed to manage uncertainty in the client’s best interests Tender documentation and the contract may be drafted to allocate uncertainty to Clarify . consequences of a change in a given factor in the chain. 5. Indirect in uences of A on B. Chains can reveal potentially significant indirect in uences of one factor on another. 6. Multiple in uences of A. be incorporated in a work package contract. Four types of management contracting system can be distinguished in the construction industry: . construction management; . management contracting; realized. But there was no point in making the project financially responsible for it with an extra £20 million in the budget. In addition to distinguishing between project management and board-level financial