INFORMS Nashville – 2016
122
2 - Vungle Inc. Improves Monetization Using Data Analytics
Ioannis Fragkos, Erasmus University, Burgemeester Oudlaan 50,
Rotterdam, Netherlands,
fragkos@rsm.nl,Bert De Reyck,
Yael S Grushka-Cockayne, Casey Lichtendahl, Hammond Guerin
Big data have enabled firms to customize their services to unprecedented levels of
granularity. In mobile advertising, once a customer enters the network, the ad-
serving decision must be made in milliseconds. In this work, we describe the
design and implementation of an algorithm we developed for Vungle Inc., one of
the largest global mobile ad networks, that incorporates machine learning
methods to make personalized ad-serving decisions. When compared to the
company’s legacy algorithm, our algorithm generated a 23% lift, which represents
a $1 million increase in monthly revenue.
MA04
101D-MCC
Low-Carbon Power Sector: Policy and
Technology Analysis
Sponsored: Energy, Natural Res & the Environment,
Energy I Electricity
Sponsored Session
Chair: Afzal Siddiqui, University College London, Department of
Statistical Science, London, WC1E 6BT, United Kingdom,
afzal.siddiqui@ucl.ac.uk1 - Strategic Offering Of A Flexible Producer
Tuomas Rintamäki, Aalto University, Espoo, Finland,
tuomas.rintamaki@aalto.fi,Afzal Siddiqui, Ahti Salo
Imbalances caused by intermittent renewable generation may give an opportunity
to a strategic producer to exert market power. We study offering strategies of a
flexible producer in day-ahead and intraday markets using a bi-level model in
which the upper-level represents the profit maximization of the producer and the
lower-level problems clear both markets sequentially. Using data from the Nordic
power market, we find that the flexible producer can increase its profit by
withholding production and by causing transmission grid congestion in both
markets. Moreover, we compare the welfare impacts of the strategies to those of
perfect competition and the dispatch policies in Morales et al. (2014).
2 - Power And Heat Market Model
Vilma Virasjoki, Aalto University, Espoo, Finland,
vilma.virasjoki@aalto.fi,Afzal Siddiqui, Behnam Zakeri, Ahti Salo
Power markets are changing, i.a. due to an increasing share of renewable energy.
This will also have effects on combined heat and power (CHP) plants and further
on the district heating (DH) sector. It is thus essential to understand the financial
and technical interrelations of these asymmetrically regulated sectors. We use
complementarity modelling to study this linkage and give a numerical example
using the Nordic energy system. We model the power system as a mix of DC and
DC load flow linearized AC lines. We formulate perfect competition and Cournot
oligopoly models and use GAMS to solve the market equilibrium. The results
provide insights e.g. into the market power impacts on CHP and DH operations.
3 - Market Power In Electricity Markets In South-east Europe
Verena Viskovic, PhD Student, UCL, 1-19 Torrington Place,
2 Cubitt Street, London, United Kingdom,
verena.viskovic.13@ucl.ac.uk, Yihsu Chen, Afzal Siddiqui,
Makoto Tanaka
We examine the effect of market power in electricity and permits markets via
single and bi-level model. We analyse potential scenarios of ownership structure
in the post-privatisation phase in South-East Europe. We expect producers with
market power to be able to influence electricity prices through permits market. In
addition, we study the effect of virtual divestitures in mitigating market power.
4 - Variability Management In Long-term Investment Models
Lina Reichenberg, Chalmers University of Technology,
lina.reichenberg@chalmers.se,Sonja Wogrin, Afzal Siddiqui
Time representation in large-scale energy investment models has been typically
governed by variability in demand. However, as CO2 abatement is becoming a
stronger driving force, variability enters also on the generation side. As a response
to this, two families of alternative time reduction methods have been developed:
one based on representative days and the other on using time slices based on
variable resources. We investigate the performance of these two families of
methods, in terms of accuracy in predicting the system plant capacity mix and
CPU time.
MA05
101E-MCC
Remuneration of Flexibility in Electricity Markets
Sponsored: Energy, Natural Res & the Environment,
Energy I Electricity
Sponsored Session
Chair: Anthony Papavasiliou, CORE, UCL, Voie du Roman Pays 34,
Louvain la Neuve, B-1348, Belgium,
tpapva@hotmail.com1 - A Revenue Adequate, Cost Recovering, Uniform Pricing Scheme
For Wind Generation
Golbon Zakeri, University of Acukland,
g.zakeri@auckland.ac.nzGeoff Pritchard, Mette Bjorndal, Endre Bjorndal
In 2010, Pritchard et. al proposed a stochastic program that would accommodate
absorbing electricity generation from wind into an electricity market. We will
present a strict improvement over this mechanism which is based on uniform
pricing, is revenue adequate in every scenario, recovers cost for each generator in
expectation, is incentive compatible and displays a number of other desirable
properties.
2 - Operating Reserve Demand Curves For Improved Pricing In
Electricity Markets With Renewable Energy
Audun Botterud, Argonne National Laboratory,
abotterud@anl.gov, Zhi Zhou, Todd Levin
We present a probabilistic method for determining operating reserve demand
curves (ORDCs) in electricity markets, accounting for the uncertainty in wind
power forecasts. We present case studies analyzing how ORDCs influence
incentives for short-term operations as well as long-term generation expansion in
electricity markets with increasing shares of renewable energy. Finally, we discuss
to what extent ORDCs reward flexible resources through improved pricing of
energy and reserves.
3 - Ramp Capability Pricing: Environmental, Economic And Reliability
Outcomes In Markets With High Penetration Of Renewables And
Flexible CCS Plants
Dalia Patino Echeverri, Assistant Professor, Duke University,
9 Circuit Drive, Box 90328, Durham, NC, 27708, United States,
dalia.patino@duke.edu, Rubenka Bandyopadhyay
Ramp Capability (RC) pricing, recently implemented by MISO, is expected to
improve economics and reliability by adequately compensating flexible ramping
resources. Coal-fired power plants retrofit with flexible Carbon Capture and
Storage (CCS) systems, would allow air emissions reductions and improved
system ramping capability. This paper explores the effects of dispatching CCS in a
market with RC products. A modified Unit Commitment/Economic Dispatch
(UC/ED) with CO2 emissions constraint and RC pricing, simulates 10-minute,
annual operations of a scaled version of the MISO power generation fleet, to
estimate changes in generators revenue, systems costs, reliability and air
emissions.
4 - Deterministic Market Designs With Efficient Scheduling Of
Flexible Ramping Products
Stefanos Delikaraoglou, Technical University of Denmark,
Elektrovej, Building 325, room 105, Kgs. Lyngby, 2800, Denmark,
stde@dtu.dk,Yves Smeers, Anthony Papavasiliou, Pierre Pinson
The variable and uncertain nature of stochastic renewables calls for revised
market designs to optimally allocate available flexibility between energy and
ramping services. Unlike stochastic dispatch models that endogenously co-
optimize these services, deterministic models require the explicit definition of
ramping products, e.g., CAISO market design. However, these products pertain
only to capacity and thus disregard the energy cost from the deployment of
flexible resources. Contrary to existing penalty-based heuristics, we propose a
systematic approach, using nested Benders decomposition, to bring the
deterministic dispatch close to the stochastic ideal in terms of costs and prices.
MA04