IEEE Power & Energy Magazine - July/August 2016 - 26

lack of common information and models in a restructured
environment. For example, bidding strategies, forward contracting, and generation expansion could, in principle, be
determined by directly modeling the equilibrium among
profit-maximizing firms. However, generation expansion
and bidding strategies are often modeled in TEP as exogenous economic uncertainties, in part because each participant of the market has different expectations and models.
Transmission expansion is directly related to management strategies focused on hedging against one or more
sources of uncertainty, sometimes in a bidirectional fashion. For example, TEP is a complement or constituent part
of market integration and resiliency, while energy efficiency
and demand response are, to some extent, supplements to
transmission investment. Also, feedback loops exist between
policies guiding TEP and TEP findings that advise policymaking. For example, regulatory measures aimed at ensuring cost recovery may be triggered by significant transmission investments needed to comply with public policy
objectives. The close relationship among strategies suggests
that TEP is part of a broader guidance process for the sustainable and efficient development of electricity markets.
The assessment of strategies is inherently multicriteria
since each strategy is affected by risks produced by multiple sources of uncertainty. For example, the integration of
regional markets can increase liquidity, competition, reliability, and resiliency, thus addressing economic, physical,
and aggregate risks (e.g., complementary energy resource
utilization, higher adequacy and more competitive electricity prices, respectively). An assessment of expansion projects and other actions must consider various performance
metrics in a multicriteria decision framework. For example,
expected energy not supplied and CO2 emissions should not
be completely monetized and added to economic measures
of costs and benefits as the security of service and emissions
reductions are objectives on their own. Other public policy
objectives also need to be assessed independently, such as
the increased integration of renewable energy sources. Thus,
plans of compromise instead of optimal ones should be
constructed in the face of multiple, potentially conflicting
objectives. Since the performance of a plan varies across
scenarios, different risk measures must be considered for
each objective and its associated performance metric.
Holistic assessments of expansion projects are needed,
considering relevant interactions to other complementary or
supplementary strategies, including different costs and benefits associated to different risks. A few management strategies
and risk or performance metrics are depicted in Figure 8, with
the purpose of illustration rather than comprehensiveness.

Optimization Approaches
to TEP Under Uncertainty
Optimization approaches to TEP under uncertainty are currently unable to deal with multiple and varied uncertainties.
Despite increasing research efforts on both methodological
july/august 2016

and conceptual contributions, no model is currently successful in simultaneously addressing short-, medium-, and
long-term uncertainties from varied sources. Hence, practical TEP studies often rely on heuristics, simplified analysis,
and manual search for solutions based on production cost
simulations, in spite of optimization models. More research
and industrial applications of optimization are required to
appropriately aid TEP decision making. Future work should
consider the main advantages and limitations of existing
modeling approaches, assessed under the conceptual framework proposed here. Three main research lines are identified as SP, robust optimization (RO), and scenario planning.
SP and RO are, to a great extent, alternative approaches
to deal primarily with medium-term and relatively wellstructured uncertainties. SP is the most powerful but scopelimited optimization tool available. It deals with Known
uncertainties for which probability distributions are completely specified for their temporal evolution. The successive
revelation of uncertainty allows SP to exploit the tradeoff
between decisions to be made here and now and wait-and-see
decisions that can be postponed to later stages. More importantly, SP can easily adopt expected utility as its objective
function, resting on von Neumann and Morgenstern's economic theory of rational choice that constitutes the "correct"
quantification and valuation of risks. However, tractability
issues counterweigh these theoretical advantages because of
the large number of discrete probability-weighted scenarios
required for SP. In contrast, RO reduces tractability issues by
modeling uncertainty by sets and not individual realizations.
It is a powerful tool for addressing some decision-making
problems involving uncertainty parameters whose true probability distribution is difficult to predict. RO's objective is
constructing solutions that are feasible against every likely
realization of uncertainty within the prescribed set, possibly
resulting in overly conservationist solutions. Although the
conservationism of RO can be controlled, the fact remains
that avoiding probabilities and a large number of discrete
scenarios comes at the price of a rather extreme valuation
of risk, since worst-case objective functions or uncertainty
realizations are considered when optimizing.
Although SP and RO may be applied to long-term
TEP, the traditional-and most implemented-approach
assumes divergent strategic scenarios for bounding longterm outcome. Instead of assuming probabilities or intervals to describe uncertainty, between three and six strategic scenarios are usually constructed to represent plausible
divergent futures, relying more on expert knowledge and
discussion than on quantitative forecasting tools. Because
of the long-term and qualitative nature of these scenarios,
it is hard to assign probabilities to each scenario, also due
to the practical difficulty of conveying consensus among
stakeholders with different and potentially conflicting
objectives and expectations. On the other hand, overconservationism of RO may be more critical in the face of
divergent scenarios.
ieee power & energy magazine

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - July/August 2016