The energy world faces unprecedented uncertainty.” This is how the International Energy Agency started its executive summary grasping the major trends and developments in the global energy relations in autumn 2010 — even before the advent of the Japanese nuclear catastrophe, the German Energiewende and Hurricane Sandy. The advent of Robert Metcalfe's “Ethernet,” the transformation from a centralized power system to a more distributed system, increased reliance on non-deterministic sources of power, customer participation in markets and the regulatory push toward sustainability are posing nontrivial challenges to the traditional utility operations in many parts of the world. Because of these shifts, changes to traditional risk management paradigms for grid reliability and stability are required.
My colleague Leonardo von Prellwitz of Cisco and I have built a risk-taxonomy for the increased complexity of emerging electric networks. We find it useful to consider four risk types that are paramount for grid planning and operations:
Randomness (aleatory) risk, associated with stochastic variations inherent in the cyber-physical electric system
Knowledge (epistemic) risk, related to a lack of knowledge (known-unknowns) about characteristics of an electric network and connected devices
Interaction risk, created by the interaction between customers, distributed energy resources, markets and elements of the electric network
Black Swan (ontological) risk, pertaining to low probability-high impact or unknown-unknowns events occurring.
The electric system has relied largely on operating margins and conservative engineering design to manage the inherent random variations on the system. Aleatory risk is rising due to increasing intermittency from variable energy resources (including wind and solar photovoltaic systems). We are also seeing increased aleatory risk on geospatial dimensions as we balance our networks through direct dispatch of distribution connected load and use interregional generation resources. It is clear that traditional deterministic methods applied separately to transmission or distribution is insufficient. Stochastic modeling and dynamic risk management techniques should be applied holistically when evaluating transmission and related distribution systems.
Likewise, the existing paradigms for power system engineering and design are inadequate for a distributed system with a high degree of epistemic risk involving the behavior of millions of independent agents/customers and the response characteristics of their generators, energy storage resources and energy management systems. Deeper situational awareness built upon an effective observability strategy using embedded sensors across the grid and customer resources will allow grid operators better understanding of grid state information to assess reliability and stability risks. However, to manage epistemic risk, situational intelligence is needed regarding the behavioral aspects of the millions of customers and/or algorithms in their devices.
Interaction risk arises from the increasing complexity of the cyber-physical grid and the transition from a roughly 40-year-old system to a 21st century electric grid. Millions of devices are being connected to the physical electric network at the same time new information and communications systems are being installed as part of smart grid efforts. The challenge is that without an effective set of architectures to guide the development, industry is actually increasing the risk associated with unintended consequences stemming from undesirable interactions. Federated decision-making across our transmission and substation SCADA, distribution management systems and optimization algorithms in customer resources will reduce the risks identified in the recent Grid2020 report from Caltech's Resnick Institute.
Black Swan-type events in the electric system seem to be occurring more frequently than expected and of great societal significance due to the human and economic consequences of sustained widespread power system failures. The challenge is two-fold: low probability “tail events” do not fit traditional engineering planning models or risk assessment tools, and true Black Swans (unknown-unknowns) cannot be quantified. In his new book Antifragile: Things That Gain from Disorder, Nassim Nicholas Taleb (author of Black Swan) attributes the rise of Black Swans to “the loss of robustness owing to complications in the design of everything.” Largely drawing on biological and cultural/political systems, he proposes antifragile risk management techniques that build upon distributed system architectures.
The electric industry will benefit from the application of stochastic-based modeling and risk assessments and antifragile architectures currently being applied in other industries such as defense, aerospace and financial markets.
Paul De Martini (email@example.com) is a visiting scholar at Caltech's Resnick Sustainability Institute and former vice president of advanced technology at Southern California Edison.