Application of System Safety
Framework in the Hybrid Socio-Technical Environment of Eurasia
Author: Azamat Abdymomunov, SDM '10, founder of the Astana
Innovation Center
Advisor: Nancy Leveson, PhD, professor of aeronautics and
astronautics and engineering systems
The Sayano-Shushenskaya
Hydroelectric Power Station is Russia’s single largest power facility and
typically produces about 24.5 billion kilowatt-hours of electricity annually. A
2009 accident, caused by the failure of a turbine and the flooding that
followed, killed 75 people and shut down the plant. This thesis examines the
hybrid characteristics that emerged in the vital sector of the planned
economy—the electricity sector—and how those characteristics contributed to the
accident.
Abdymomunov, a former strategic
advisor to the prime minister of the Republic of Kazakhstan, examines some of
the root causes of the disaster. He notes that 20 years after splintering into
15 national entities, the Soviet region’s political and social transformation
has led to hybrid structures in political, economic, and technological domains.
Different institutions and policies, state and private business entities, old
and new technologies, and collectivist and individualist value systems coexist.
The roles of government, state enterprise, private business, and civil society
are not clearly defined.
He argues that the traditional
approach to explaining the accident—characterized by a culture of blame, as
well as a focus on reliability, responsibility, and modernization—fails to
acknowledge systemic causes. The borders of systems change over time, evolving
and adjusting to the external environment. In the case of the power station,
the reform of Russia’s electricity sector affected station safety and
reliability.
Abdymomunov concludes that accidents are complex processes involving the entire sociotechnical system. In the case of the Russian power plant, insufficient capital investment and backlog in maintenance shifts were key systemic factors that allowed organizational behavior to migrate from a safe to an unsafe state.
A Systems Approach to Food Accident
Analysis
Author: John Helferich, SDM ‘10, adjunct professor at the College
of Business Administration at Northeastern University and a Batten Fellow at
the Darden School of the University of Virginia
Advisor: Nancy Leveson, PhD, professor of aeronautics and
astronautics and engineering systems
Food-borne illnesses lead to
3,000 deaths per year in the United States. Some industries, such as aviation,
have significantly increased safety through careful accident analyses and
follow-up changes in industry practices. In the food industry, however, current
methods of accident analysis remain grounded in regulations developed when the
industry was far simpler than it is today.
Helferich, the former senior
vice president of research and development at Mars Incorporated, explores
whether the incidence of food-borne illness could be reduced by changing the
method of accident analysis from the current approach to a system theoretic
method. Currently, food accident analysis combines epidemiology—to identify and
track illnesses to their origin—with a regulatory standards approach that is
based on a linear model of accident causation.
Helferich asserts that causal
analysis using system theoretic accident modeling processes (CAST) is a better
method. CAST is designed to determine how and why accidents occurred by
analyzing the control structure of the accident system.
To test whether CAST can
provide more comprehensive insights into food system accident causation,
Helferich analyzed a case study using both traditional and CAST approaches. The
case study chosen is a 2008 outbreak of salmonella linked to the Peanut
Corporation of America, which forced the company out of business and
significantly drove down sales of peanut butter nationwide
Helferich demonstrates that the
application of a system theoretic accident analysis method such as CAST results
in more learning than the current method of accident analysis. He concludes
that the current food accident analysis system should be augmented with a
systems-based approach to identify and control more extensive system hazards.
Platform Project
Management: Optimizing Product Development by Actively Managing Commonality
Author: Raúl Pinillos Montaño, SDM ‘09, program management engineer
at Ford Motor Company
Advisor: Olivier de Weck, PhD, associate professor of aeronautics
and astronautics and engineering systems
Product developers continually
strive to create better products faster and more cheaply. One way to do that
has been to design common parts and create product families, which enables
companies to increase profitability by sharing costs among different products.
However, managing common designs in a product family is not a trivial task.
Over time, product commonality typically decreases, a phenomenon called
divergence.
Pinillos proposes a system
dynamics methodology to explain the different impacts of divergence on the
development projects within a product family. Using a model calibrated via a
case study in the automotive industry, he simultaneously simulated lead and
derivative projects based on product commonality.
Pinillos measured divergence
rates and found they ranged from 0.4 percent to 1.2 percent loss of product
commonality every month. Further, he found that this divergence caused
significant damage to the product development projects—as high as a 22 percent
schedule overrun or a 29 percent increase in personnel needed to achieve the
planned project schedule. He reports that these effects grew proportionally to
the divergence rate.
Pinillos concludes that
actively managing product commonality can be an effective method to achieve a
successful execution of the development projects when the product platform
approach is utilized. Platform project management, he argues, can ensure that
product commonality is achieved throughout the development process by balancing
and trading off all the designs among the entire product family rather than by
making decisions for each product in isolation.



No comments:
Post a Comment