Managing Shortcuts in Engineering Systems and Their Safety Effects: A Management Science Perspective (PSAM-0036)

Garber, Russ G.; Paté-Cornell, Elisabeth

doi:10.1115/1.802442.paper251

The issue of corner cutting has vital implications for the safety of complex engineered systems because shortcuts can lead to unacceptable (and unknown) technical failure risks. In addition, corner cutting is often illicit. Therefore, the system's managers are not necessarily aware of it and may be basing their decisions on wrong assumptions about operators' or technicians' actions. Illicit shortcuts occur for a variety of reasons, among them unrealistic deadlines and resource constraints, bad incentive structures, divergence of risk attitudes, ignorance of the system, myopic views, and wrong learning from past decisions and outcomes.

This paper addresses the issue of corner cutting in the development, management, or operations phases of a complex engineered system. The purpose of this model is to provide decision support to managers in situations where an agent/operator may be tempted to take shortcuts. The principal-agent format is used to model the situation, the “corner-cutter” being the agent, and the supervisor the principal.

Both the principal P (“she”) and the agent A (“he”) are assumed to make choices that maximize their expected utilities. Two key equations determine their choices (assumed to be rational) and the interactions between them. The agent's decision is to take a particular course of action, k*^A, among the set of options {k^A} that he believes to be available to him (including some not allowed by the principal). Let i be the index of the outcome scenario for the system (e.g., failures or not) given his actions (e.g., short cuts). Let c^p be the constraints and In^p the incentives, both set by the principal. Note (p_i^A|k^A) as the agent's subjective probability of scenario i, conditioned upon the option that he chooses. His utility for each scenario i is determined by its consequences to him, given the incentives and constraints. His best option k*^A is thus determined by:

{E U}^{A} (k^{* A}) = \underset{{k^{A}}}{Max} \sum_{i} [(p_{i}^{A} | k^{A}) \times U^{A} ({consequences}^{A} ({scenario}_{i}) | c^{P}, {I n}^{P})]

The principal controls some of the factors (shown in the equation above) that determine the agent's expected utility by setting and communicating to him both the constraints and the consequences to him associated with each outcome scenario i. The principal also influences the available alternatives through the agent's training and the information given to him about the system. She can also have an effect upon the agent's preferences through hiring practices and a corporate safety culture. The principal can thus evaluate the probability that the agent takes a course of action k_j^A in response to both the constraints and the incentive system that she sets, and she will choose those (noted c*^P and In*^P) among her possible options that maximize her expected utility over all possible system outcomes and their consequences to her:

{E U}^{P} (c^{* P}, {I n}^{* P}) = \underset{{c^{P}, {I n}^{P}}}{Max} \sum_{j} [(p_{k_{j}^{A}}^{P} | c^{P}, {I n}^{P}) \sum_{i} (p_{i}^{P} | k_{j}^{A}) \times U^{P} ({consequences}^{P} ({scenario}_{i}))]

In this paper, we analyze the interaction between the decisions of the principal and the agent using these two equations in an interactive mode. One of the key features of the model is the use of probabilistic risk analysis to determine the effect of the agent's action (and shortcuts) on the performance of the corresponding component or subsystem, and therefore on reliability of the whole system. The overall model then allows us to quantify the effects of two key decisions of the principal (constraints and incentives) on the actions of the agent, and consequently, on the performance of the system. This model can thus support the managers' decisions based on anticipating their consequences on the system's performance instead of simply setting rules without considering their effects.

251 Managing Shortcuts in Engineering Systems and Their Safety Effects: A Management Science Perspective (PSAM-0036)

Sign In

Purchase this Content

Email alerts

ASME Journals

ASME Conference Proceedings

ASME eBooks

Resources

Opportunities

251 Managing Shortcuts in Engineering Systems and Their Safety Effects: A Management Science Perspective (PSAM-0036)

Sign In

Purchase this Content

Product added to cart.

Email alerts

Related Chapters

Related Articles

Related Proceedings Papers

ASME Journals

ASME Conference Proceedings

ASME eBooks

Resources

Opportunities

This Feature Is Available To Subscribers Only