Bilevel Optimization Framework For Smart Building-To-Grid Systems

This article has been accepted for publication in a future issue of this journal, but has not been
fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2016.2557334, IEEE
Transactions on Smart Grid
IEEE TRANSACTIONS ON SMART GRID, VOL. , NO. , 1
Bilevel Optimization Framework for

Smart Building-to-Grid Systems
Meysam Razmara, Guna R. Bharati, Mahdi Shahbakhti, Sumit Paudyal, and Rush D. Robinett III
Abstract—This paper proposes a novel framework suitable electricity price. Increased load at those hours may impact
for bilevel optimization in a system of commercial buildings voltage performance in the distribution grid. In [9], it is
integrated to smart distribution grid. The proposed optimization demonstrated that uncontrolled penetration of electric vehicle
framework consists of comprehensive mathematical models of
commercial buildings and underlying distribution grid, their loads violates the voltage standards set by the ANSI. Thus,
operational constraints, and a bilevel solution approach which it is advisable that DSM should be employed considering
is based on the information exchange between the two levels. the operational requirements of customers as well as the
The proposed framework benefits both entities involved in electric power grid. In DSM activities, distribution utilities are
the building-to-grid (B2G) system, i.e., the operations of the more concerned about feeder loss minimization, load factor
buildings and the distribution grid. The framework achieves two
distinct objectives: increased load penetration by maximizing improvement, reactive power optimization, etc., as part of
the distribution system load factor and reduced energy cost their operational objectives [10]–[15]. On the other hand, the
for the buildings. This study also proposes a novel B2G index, objectives of DSM activities at customer level are mainly
which is based on building’s energy cost and nodal load factor, focused on comfort maximization and cost minimization. This
and represents a metric of combined optimal operations of illustrates that the objective set by the grid and customers in
the commercial buildings and distribution grid. The usefulness
of the proposed framework is demonstrated in a B2G system DSM activities could often be conflicting.
that consists of several commercial buildings connected to a In [16], [17], promising results are obtained from DSM in
33-node distribution test feeder, where the building parameters reducing customer’s energy price. In [16], pre-cooling and pre-
are obtained from actual measurements at an office building at floating is performed to reduce total electricity cost. In [17],
Michigan Technological University. optimization is performed with a trade-off between cost and
user comfort. In [8], HAN is employed to reshape the demand
Index Terms—Building to grid optimization, smart building, profile based on the impact on distribution transformer. In
smart grid, HVAC, model predictive control, hierarchical opti-
DSM models proposed in [8], [16], [17], customers obtain
mization, demand response, load shedding.
benefits by shifting the loads. The mathematical models used
in [8], [16], [17] consider peak power information from
I. INTRODUCTION distribution grid as one of the constraints of optimization but
B UILDING sector in the United States accounts for

about 70% of electricity energy consumption [1], in
which 41.4% of energy consumption is directly related to
detailed modeling and operational constraints of distribution
system have not been considered. Thus, for advanced DSM
activities in Smart Grids, detailed mathematical modeling
the space heating, ventilation, and air conditioning (HVAC). considering objectives and operational requirements of cus-
Thus, HVAC systems have a great potential to reduce the tomers and the distribution system with real-time information
energy usage/cost in buildings. As reported in [2], [3], smart exchange between customer and the grid is crucial; which is
control techniques, such as model predictive control (MPC), a major focus of the proposed work.
can provide significant energy saving. Development of MPC Fig. 1 summarizes past studies related to three research
for optimal control of HVAC systems has been a focus in [4], categories: building, grid, and integrated building-grid opti-
[5], which yields benefits to the operations of buildings and mization. In the first category, optimal and model predictive
electric power grid. unidirectional building optimization are studied [2]–[5], [18]–
Smart grid technologies, such as smart appliances and home [27]. In this category, building’s HVAC performance or opera-
area network (HAN), render the electrical loads more man- tional cost is optimized using optimal or model predictive con-
ageable and controllable at customers premise, and facilitates trol technique. The second category include studies in which
advanced demand side management (DSM) activities [6]. distribution grid operation is optimized for various operational
However, existing mathematical models developed for cus- objectives considering aggregated loads [10]–[15]. The third
tomer side DSM mainly consider minimization of electricity category includes building-to-grid (B2G) integration in which
cost [7], [8], which results in increased load at hours with low performance of buildings including comfort level and energy
cost is optimized considering grid side information [7], [16],
M. Razmara, M. Shahbakhti, and R. D. Robinett III are with the Depart-
ment of Mechanical Engineering-Engineering Mechanics; G. R. Bharati and [28]–[33], and demonstrates its usefulness for grid services.
S. Paudyal are with Department of Electrical and Computer Engineering, However, in the modeling in [7], [16], [28]–[33] distribution
Michigan Technological University, Houghton, MI, USA. Emails: {mrazmara, grid is simplified or completely ignored. Survey results in
grbharati, mahdish, sumitp, rdrobine}@mtu.edu.
Fig. 1 reveal that there are extensive works accomplished in the
area of control/optimization of distribution grid and buildings
1949-3053 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2016.2557334, IEEE
bidirectional information exchange cloud computing can be

deployed [35]. VOLTTRON, OpenADR, BEMOSS are some
of the open source platforms that can be utilized for the
implementation of the proposed models at the customer and
grid levels [36]–[38].
Fig. 1. Summary of past studies on building, grid, and B2G optimization.
HVAC systems independently. However, there is little work

done in the area of bi-directional B2G, including mathematical
modeling required for the B2G integration. This work proposes
a bi-directional B2G optimization framework based on detailed
mathematical modeling of a B2G system.
The proposed B2G framework is based on information ex-
change between the two levels, i.e, the BEMS and distribution Fig. 2. Conceptual building to grid communication framework.
system operation (DSO) control center. The predicted and
optimized load profile of buildings are provided to the DSO by This study improves previous single level optimization of
the BEMSs. Similarly, maximum allowable load penetration buildings by considering the grid model and builds upon
information, that ensures feasible grid operations, are provided preliminary results in [39] by the authors. To our knowledge,
to the BEMSs by the DSO. it is the first study that proposes a bilevel optimization
For the demand response applications discussed in this framework in B2G interaction, which benefits both building
work, a bidirectional communication infrastructure is required, and grid operations. The contribution of this paper is on the
as shown in Fig. 2, between the BEMSs and the DSO. Also, development of generic hierarchical optimization framework
unidirectional communication links are required between the for B2G system, which is essential for coordinated control
BEMSs and the market operator (MO), BEMSs and local of multiple BEMSs connected to distribution grid for large
weather station, DSO and control equipment at distribution scale demand response and other grid level services. More
level. The required communication infrastructure must be specifically, to reach this goal this paper:
secure, reliable, and low-cost for autonomous interactions [34], 1) develops a physics-based comprehensive mathematical
[35]. In this proposed work, the interaction between the two model of HVAC system and model predictive controller
levels, i.e., the building and grid, are kept at the minimum; (MPC) with the aim of minimizing buildings electricity
thus, the communication bandwidth usage is sporadic. The costs. The MPC controller works in grid friendly man-
information that building receives from the grid control center ner, i.e, it communicates with distribution grid control
is the maximum demand limit, which can be sent once every center and incorporates the constraints set by the grid
15-30 minutes for next couple hours to next day. Building controller required for the feasibility of grid operation.
receives information from utility or MO about energy price 2) develops a detailed and generic mathematical model of
once per day (i.e., the day ahead energy price) and/or every distribution grid that coordinates with several building
5 minutes to one hour (i.e., the real-time energy price), controllers to optimize the operation of the power grid.
depending on which energy rate is applied to the building This paper proposes a novel B2G index that ensures the
customers. Building receives weather forecast on daily basis benefits at the grid level from the building side optimization,
and/or the hourly weather update. It should be noted that the and develops a coordination algorithm to solve the hierarchical
information flow depends on the variation of the parameters in B2G framework.
the building and grid models during the course of operation, Rest of the paper is organized as follows. Section II
and the type of services the building loads provide to the grid presents mathematical models of building and distribution
(such as demand response, regulation). The communication system, MPC based optimization model at building level,
infrastructure must be flexible to interact with the existing and distribution grid optimization model. Section III discusses
and future BEMSs and communication protocols such as methodology for bidirectional optimization and the proposed
Modbus, RS-484, BACnet, etc. [34]. To manage real-time and B2G index. Section IV describes the building testbed and dis-
tribution test feeder. The results of case studies are presented Function g is approximated as a linear function of Tkr (t),
in Section V, and the main conclusions from this work are Qrad int r
i (t) and Q̇i (t). Thus, the disturbance is w = aTk (t) +
rad int
included in Section VI. bQi (t) + cQ̇i (t).
Heat transfer of each wall and room equations and distur-
bance form the system dynamics of building are represented
II. M ATHEMATICAL M ODELING
in state-space form by:
A. Building Components Thermal Modeling ẋ(t) = f (x(t), u(t), w(t), t)
Convection, radiation and conduction are the main heat y(t) = Cx(t) (4)
transfer ways in buildings which are time varying. We use n
where x(t) ∈ R is the state vector. State vector includes
a common building modeling approach known as nodal ap-
nodes’ temperature in the nodal network. y(t) ∈ Rm is
proach to model building’s construction and materials with
the output vector of the system presenting thermal zones’
electrical components such as resistors, capacitors, and current
temperature. Thermal zone is a section of a building that is
sources [2], [40]. For this purpose, buildings are considered
required to meet certain comfort level. In this paper, each room
as graphs containing walls and rooms as nodes. p represents
is considered as a thermal zone. The input vector which is the
number of nodes, q out of p nodes are rooms, and the
air mass flow rate and its temperature for each zone is shown
remaining p − q nodes are walls. i = 1, . . . , q numbers are
with u(t) ∈ Rl×m . l is the number of inputs to each thermal
assigned to each room. The following equation determines the
zone (i.e., air mass flow and supply air temperature). C is the
heat transfer of the walls [40]:
output matrix with proper dimension to return output, y(t),
w X Tkr − Ti,jw
dTi,j 1 from states.
= w ( wi,j + ri,j γi,j Aw rad
i,j Qi,j ) (1)
dt Ci,j w Rk
In our testbed, HVAC air mass flow rate is directly measured
k∈N i,j
using a vent mass flow meter. Density, heat capacity, trasns-
where Ti,jw
is the temperature of the wall between room i and missivity and etc. are determined using standard engineering
room j, Ci,jw
is the heat capacity of the wall between room i thermodynamics tables. There are unknown (e.g., wall heat
and j, Tkr is the temperature of adjacent rooms to wall wi,j . capacitance and outside air convection coefficient) that are
Thermal resistance between the center-line of wall (node wi,j ) determined for the testbed using Unscented Kalman Filter
w (UKF) techniques presented in [2].
and the neighboring node k is denoted with Rk i,j . γi,j and
Aw Equation (4) describes the nonlinear time evolution of the
i,j are radiation heat absorption coefficient and area of wall
between room i and j, respectively. Wall identifier is shown system. The supply air temperature (i.e., input to the system)
by ri,j which is equal to 0 for internal walls, and equal to is multiplied by the air mass flow, which is the other time-
1 for peripheral walls. Qrad varying known input to the model. Given that the air mass
i,j represents the radiative heat flux
density on wall (i, j) while Ni,j w
is the set of all neighboring flow rate is constant during day, we can consider the linearized
nodes to node wi,j . form of the system. We use Euler’s discretization method to
The following equation governs the temperature of the ith discretize the state update equation (4) for controller design.
room [40]: The state-space model is given by:
dTir 1 X h Tk − Tir i xk+1 = Axk + Buk + Ewk (5)

= r( + πi,k τkw Awin
i,k Q rad
i
dt Ci Ri,k (2) where wk is the disturbance vector at instant k and E is the
k∈Nir
disturbance coefficient matrix with appropriate dimension.
+ ṁri cpavg (Tis − Tir ) + Q̇int
i ) Fig. 3 represents the experimental validation of the building
where Cir and ṁri denote the heat capacity and air mass flow thermal model for few days in winter. Fig. 3(a) compares
into or out of the room i, respectively. Tk is the temperature the simulated and measured room temperature, and Fig. 3(b)
of surrounding node k to room i. cpavg denotes the average illustrates the heat-pump power based on the difference be-
specific heat capacity of air and Tis is the temperature of the tween the room temperature and the measured supply air
supply air to room i. πi,j is window identifier which is equal temperature. Details of HVAC heat-pump system are discussed
to 0 if there is no wall between room i and j, otherwise in Section IV.
equal to 1. Awin
i,j is the total area of window between room i
w
and surrounding room j, τi,j is the transmissivity of glass of B. Building Optimization Model-I
window between room i and j, Qrad i is the radiative heat flux The main goal of optimization is to minimize energy cost
density per unit area radiated to room i, and Q̇inti denotes the by the building’s HVAC system. For this purpose, both HVAC
internal heat generation in room i. Nir is the set of all nodes energy usage and time varying electricity price are taken into
surrounding room i. account. Energy usage is characterized by the energy index Ie
The disturbance to the model, w(t), is function of neighbor- which is defined as [2]:
ing room temperature, Tkr (t), internal heat generation in rooms 24
X
Q̇int rad
i (t), and radiative heat flux density on walls, Qi (t). The Ite = PtB ∆t (6a)
disturbance is given by [40]: t=1
w(t) = g(Tkr (t), Qrad int PtB = Pth + Ptc + Ptf + Pto (6b)
i (t), Q̇i (t)) (3)
(a) output constraint on temperature of room and (8g) represents

Temperature (oC) the constraint on slack variables. Constraints (8b) and (8d)
21
Room air
20 should hold for all k = 0, 1, ..., N − 1, and N is the prediction

19
o
STD of error: 0.09 C Measured horizon. In section V-F, effect of N on simulation results
Mean of error: 0.1 oC Estimated
will be discussed. (8c), (8e), (8f), and (8g) should hold
18
MON TUE WED THR for all k = 1, 2, ..., N . Ut = [ut|t , ut+1|t , · · · , ut+N −1|t ]
(b) represents control inputs vector and ut+1|t is the estimated
1000
Measurerd value of ut+1 at time t, ǫt = [εt+1|t , · · · , εt+N |t ] and ǫt =
Heat−pump
power (W)
STD of error: 14.6 W

Mean of error: 16.7 W
Estimated
[εt+1|t , · · · , εt+N |t ] are the slack variables. Slack variables
500
are added to ensure feasibility of optimal control problem. In
equation (8a) there is penalty (ρ) for slack variables. Therefore,
0
MON TUE WED THR by choosing large values for ρ, the optimizer enforces the
Time (day) slack variables to take small values and does not violate the
Fig. 3. Experimental validation of the building thermal model for a sample temperature constraints. ρ characterizes the flexibility of a
room/zone.
building to provide services at the grid level. ǫ is not a pre-
determined value in the optimization; however, by choosing
where PtB is the active power consumption of building. P h is large enough values for ρ, deviation from comfort bound (i.e.,
heating power, P c , and P f are cooling power and fan power, ǫ) is guaranteed to be minimum. In Equation (8a), lower values
respectively. We denote all buildings’ other loads with P o for ρ leads to higher flexibility for grid level services; however,
which includes lighting and appliances loads. Heating, cooling this results in occupants’ discomfort [2]. Equation (8f) defines
and fan power consumption are determined by [2]: the relationship between the room temperature and the slack
Pth = ṁri (t)cp,air [Tth − Ti,t
r
] (7a) variable value. Ω is dynamic pricing of electricity which is
considered to be independent of power consumption by the
Ptc = ṁri (t)cp,air [Ti,t
r
− Ttc ] (7b) building loads. yt+k|t is the vector of thermal zone temper-
Ptf = λ(ṁri )3 (7c) ature, dt+k|t is the disturbance load, and T t+k|t and T t+k|t
for k = 1, · · · , N are the lower and upper bounds on the
where T h and T c are the supply air temperatures in the
zone comfort level, respectively. U t+k|t and U are the lower
heating mode and the cooling mode, respectively and these
and upper limits on the supply air temperature delivered by
variables are the system control inputs. Equations (7a) and (7b)
the heat-pump of the HVAC system, respectively. Operational
represent the relationship between the HVAC heat-pump power
limit on maximum supply air temperature is not time varying,
consumption and the temperature difference. λ [W.s3 .kg−3 ] is
therefore time invariant constraint U is used. δU and δU
the coefficient of fan which defines the cubic relation between
are limitation on rate of change of supply air temperature
power required and mass flow rate. Since air mass flow rate is
according to dynamics of heat-pump condenser.
not a control input in the state model, and it is constant during
the daytime, fan power, P f , does not affect optimization result.
The temperature difference between supply air and the C. Building Optimization Model-II
room air is proportional to electrical energy consumption. Objective of this optimization model is the same as (8).
Hence, the building controller keeps the room temperature However, an additional constraint is used as a feedback from
within the comfort levels such that energy cost is minimized. distribution grid, which is given as:
We use the proposed objective function in equation (8a) to
PtB ≤ Ptvar (9)
minimize energy cost. In addition, soft constraints (i.e., ǫ) are
implemented to guarantee feasibility of optimal solution at all where Ptvar is the maximum allowable building active power
times. The following optimization problem is being solved at penetration in the distribution grid, which will be discussed
each time step t, and cumulative cost is calculated from t to in Section II-D. Equations (6), (7a)-(7c), (8b)-(8g), and (9)
tmax : tX
max
define equality and inequality constraints of the optimization
min {( Ite ΩT ) + ρ(|ǫt |1 + |ǫt |1 )} (8a) model.
Ut ,ǭ,ǫ
t
subject to: D. Distribution Grid Optimization Model-I
xt+k+1|t = Axt+k|t + But+k|t + Edt+k|t (8b) Objective of this optimization model is to determine max-
imum additional building loads that can be connected at
yt+k|t = Cxt+k|t (8c)
different nodes of the distribution system. This information
U t+k|t ≤ ut+k|t ≤ U (8d) is used as a feedback to the building optimization model-
δU ≤ ut+k+1|t − ut+k|t ≤ δU (8e) II described in Section II-C, which represents information
T t+k|t − εt+k|t ≤ yt+k|t ≤ T t+k|t + εt+k|t (8f) exchange between the two levels of the bilevel optimization
framework. Mathematically, objective function can be written
εt+k|t , εt+k|t ≥ 0 (8g) as: (t )
max X
X
where (8b) and (8c) form building’s state model, (8d) and max var,nb
Pt (10)
(8e) are constraints on input i.e. supply air temperature, (8f) is t=1 nb∈n
where n represents nodes and nb represents nodes where In Equations (14a, 14b, 14c, & 14d) overbar and underbar
buildings are connected. denote the maximum and minimum value for variable, respec-
Equality constraints of the grid optimization model include tively. Equation (14a) ensures that the voltage limits are within
component modeling and voltage/current balance equations. the standard set by ANSI. Equations (14b) and (14c) are the
Distribution feeders, transformers, and voltage regulators with limits on tap positions in transformer and capacitor banks,
load tap changers (LTCs) are modeled using ABCD param- which take integer numbers only.
eters as described in [41]. Constant power, constant current,
and constant impedance loads are considered at each node. E. Distribution Grid Optimization Model-II
Mathematically, these equality constraints can be represented Objective in this optimization model is to maximize the
as: system load factor. Mathematically, this objective function is
m
Vtn a bm Vtn+1 written as [42], [43]:
= m (11a)
Itsen,m c dm Itres,m (P
tmax
)
t=1 Pt
max (15)
tmax |Pt |∞
Itres,m−1 = Itsen,m + ItZ,n + ItI,n
where, Pt total active power load connected to the distribu-
+ ItP,n + ItC,n + Itvar,n (11b)
tion grid at hour t is calculated as:
Z n ItC,n = Vtn Ctapnt (11c) X n o X
Pt = ℜ Vtn (ItZ,n + ItI,n + ItP,n )∗ + Ptvar,nb
where m represents feeder branches, Itres,m−1 and Itsen,m n nb
are receiving and sending end currents on branch m − 1 and (16)
m, respectively. For branches with LTC, am = 1+∆s·T 1
, Equations (11), (12), and (14) represent the equality and
tapm
t
−1
bj = cj = 0 and dj = aj where ∆s represents voltage inequality constraints. Load factor in (15) is improved by
change (p.u.) with one step change of transformer tap position controlling LTC and switched capacitor banks. For grid power
and T tapm n flow analysis and distribution grid optimization model-II,
t represents LTC tap positions, Z is the impedance
of single capacitor in a bank at nominal power and voltage. Ptvar,nb = PtB,nb , which is obtained from (6).
ItP,n , ItI,n , ItZ,n , and ItC,n are the currents from the constant
power, constant current, constant impedance loads, and ca- F. B2G Index
pacitor banks connected at node n, respectively. Vtn represents In the hierarchical approach to solve the proposed optimiza-
nodal voltage and Ctapnt represents number of capacitor banks tion models, the lower level is concerned about the electricity
switched on. costs in buildings, while the upper level is concerned in maxi-
Base loads in the distribution feeder are modeled using sum mizing the system load factor. A worthwhile B2G optimization
of constant power, constant current, and constant impedance should maintain an appropriate compromise between low elec-
loads. Building loads are additional power to the grid and tricity cost in buildings and high load factor in the distribution
modeled as constant power load. Building load current in system. To account for this, we define a new B2G index (IB2G )
terms of building power can be calculated using: to assess the performance of the proposed B2G optimization
framework, which allows reduction in the electricity costs and
Ptvar,nb
Vtnb (Itvar,nb )∗ = ∠φ (12) improvement of system load factor starting at the building level
cosφ with the use of nodal load factor. The B2G index is defined
where φ is the power factor angle of the building load. An as:
additional equation is needed to ensure that building load
−1
Lf
z }| {
penetration is allowed fairly in the distribution grid. Thus, a B,nb
tmax Pt (17)
fairness index (ItF ) is defined to ensure fair distribution of
IB2G = α (Ite ΩT ) + β Ptmax B,nb∞
building loads as:
t=1 Pt
P var,nb
ItF = t E,nb (13) where α and β are the weight factors for energy price (Ie ΩT )
Pt
and nodal load factor (Lf ), respectively. The smaller the IB2G
where, PtE,nb is the base load in nodes nb. index, the better the performance for the B2G optimization.
Inequality constraints of the distribution grid optimization The ratio of weight factors, α to β, determines the importance
model include limits of voltage as per ANSI standards, limits of energy cost compared to the system load factor. α and β
on capacitor and LTC positions, limits on ItF , and other limits are the design parameters in the B2G optimization problem
such as feeder capacity, transformer capacity, etc. Mathemat- and can be adjusted by the building and grid operators,
ically, the inequality constraints can be written as, depending on whether the benefits from the building side or the
benefits from the grid side are more desirable. Thus, selection
V ≤ Vtn ≤ V (14a)
n
of proper values of α and β for optimal operation system
0≤ Ctapnt ≤ Ctap , ∀ Ctapnt ∈I (14b) requires sensitivity analysis of the building-grid system under
T tap ≤ T tapmt ≤ T tap, ∀ T tapm
t ∈I (14c) consideration.
F F In the case studies, IB2G is used as an alternative objective
0 ≤ It ≤ I (14d) function to (8a) in building optimization models. Note that
(8a) does not account for grid’s objective; while the proposed
B2G index considers the grid’s objective in the building
optimization model with consideration of the nodal load factor.
III. S OLUTION M ETHOD

A solution method, shown in Fig. 4, is proposed for the
bilevel B2G optimization framework. The proposed solution
method consists of information exchange between the two
levels, i.e., the commercial buildings and the grid control
center; thus, it is bidirectional in nature. The B2G optimization
framework consists of two optimization models for the demand
side (i.e., building systems) and two optimization models for
the supply side (i.e., distribution grid). These optimization
procedures include:
• Building Optimization Model-I: Building energy cost
minimization;
• Distribution Grid Optimization Model-I: Maximizing
penetration of building loads in the distribution grid;
• Building Optimization Model-II: Minimizing building
electricity cost considering constraints from grid side (9);
• Distribution Grid Optimization Model-II: Maximizing
load factor of distribution grid.
The solution procedure begins with the demand side opti-
mization with an objective to reduce electricity cost for the
next day (label A in Fig. 4). At this stage, the optimized
load profiles of each of the commercial buildings are made
available to the distribution grid control center. A distribution Fig. 4. Proposed bidirectional B2G optimization flowchart.
grid power flow analysis is carried out to confirm the feasibility

of building load profiles. The feasibility test checks various op-
to handle this issue is to consider a small unused margin
erational constraints of the distribution grid including voltage
of power in the maximum allowable loads in (10) that can
limits as per ANSI standards defined by (14a)-(14c). Violation
accommodate change in power consumption of loads due to
of operational constraints makes the building loads in-feasible
the voltage change. However, this change will not affect the
for grid operation. In case the building loads are feasible for
feasibility of HVAC system since they can be considered as
the grid (label B in Fig. 4), tap positions of capacitor banks and
constant power load [44].
transformers are calculated using distribution grid optimization
model-II, which maximizes the system load factor.
In case of in-feasibility of the building loads from grid IV. T EST S YSTEMS
B,ib
side, the maximum allowable load, P t , is obtained from
B,ib A. Building Testbed
the distribution grid optimization model-I. P t is used as a
feedback to the BMS system (label D in Fig. 4) to solve the A commercial three story Lakeshore Center building with
building optimization model-II. If the updated optimization an area of 61, 500f t2 at Michigan Technological University
is feasible, then the optimized load profile is sent back to (MTU), Houghton, Michigan, is considered as the testbed. The
the distribution optimization model-II to maximize load factor testbed is equipped with ground-source heat-pumps (GSHP)
(label F in Fig. 4). Otherwise, other options like chang- to provide required energy for heating and cooling. GSHPs
ing building’s temperature requirements (label E in Fig. 4) transfer geothermal energy from the ground to rooms. GSHPs
becomes necessary. Next, building optimization model-II is are HVAC energy efficient technologies with high coefficient
solved by considering the feedback from the grid as additional of performance (COP). The GSHPs in this study have a
constraint. Then, the resulting building load profile information nominal COP of 3.2.
is sent to the distribution grid control center, which is used to Each room is equipped with a GSHP to maintain comfort
solve distribution optimization model-II. of occupants. Fig. 5 shows the schematic of the building
Change of the control variables, i.e., LTC and capacitor with GSHPs connected to the distribution grid at node #10.
banks may result in different node voltages from the two Temperature of each zone in Lakeshore Center building is
distribution grid optimization models, which in turn can impact measured using a temperature sensor from the BMS with
the voltage dependent loads and make the grid optimization accuracy of ±0.2◦ C. HVAC system in the testbed consumes
problems in-feasible. Thus, depending on the types of load up to 44% of electricity in winter, while the rest of electricity
used in the distribution grid optimization model, the whole consumption is attributed to lighting, computers, and office
solution process may need to reiterate. An alternate approach appliances. Historical data of energy consumption of an office
building at MTU is used to compute the share of energy Weather forecast information available from a local weather
consumption of major loads in the testbed. station is used. The constraints on the room temperature are
set based on ASHRAE standards [48]. Other building and grid
parameters used in the simulation can be found in [20] and
[11], respectively.
The optimization results are compared with the non-
optimized rule-based controller (RBC). We use same ∆t = 1
hour time step to have a fair comparison among B2G optimiz-
ers in this study. Given the slow thermal behavior of the testbed
building and the type of demand response applications, one
hour time step is sufficient to capture the dynamics. Depending
on building type, a shorter time step may be required.
For the case study, five buildings each having twenty zones
are connected at node 2 of the distribution grid, similar
three buildings are connected at node 10, four buildings are
connected at node 18, and seven buildings are connected
at node 31. In this paper, the optimization results are only
presented for the buildings at node #10 (see Fig. 5). The nodal
load factor is presented for the node #10 as a node representing
Fig. 5. Schematic of the testbed with a ground-source heat-pump. Two B2G system. In addition, system level load factors are also
temperature sensors are used to measure the average room air temperature. determined for the distribution grid system to compare the
operation of different B2G optimizers.
As mentioned in the Section I, HVAC systems account for a
B. Distribution Test Feeder large amount of load in buildings. Therefore, in this study, we
focus on controlling the HVAC load of the buildings which is
A standard single phase 12.66 kV, 33-node distribution
considered to be dispatchable. Other loads in buildings includ-
feeder [11] is considered for the case studies. The standard test
ing lighting, appliances, and base loads in the distribution grid
system is modified by adding a regulator and capacitor banks
are considered to be non-dispatchable. For MPC simulation, a
to account for the control equipment available in distribution
prediction horizon of N = 24 is used with time step of 1 hour.
grids.
Here, we introduce three different optimization methods and
As shown in Fig. 5, a transformer is equipped with voltage
compare them with the RBC.
regulator (connected between 10th and 11th nodes in the test
systems), which maintains the voltage at node 11 within ±10%
of the nominal value. A 32-step regulator is considered, and a A. Unoptimized Rule Based Control
single tap change represents voltage of 0.00625 p.u. Capacitor In this approach, when temperature goes below the lower-
banks are connected at 8th and 16th nodes of the feeder and bound temperature, the HVAC RBC keeps the heat-pump
are modeled as constant impedance load capable of delivering compressor on for the duration of ∆t. In the next time step,
reactive power to the grid. Identical five units of capacitors the RBC checks the room temperature again and determines
with 10 kVAr each are considered, which are represented as whether the room temperature is within the comfort bounds.
switched capacitors with tap positions from 0 to 5. The monthly electricity cost of the building and energy con-
It is assumed that smart buildings are connected at four sumption along with other B2G metrics are listed in TABLE I.
arbitrary nodes, i.e., nodes 2, 10, 18, and 31. Each building
is considered to have twenty identical zones with similar load B. Building/Grid side Optimizations
profiles and with the same temperature requirements.
The results of uncoordinated building side and grid side
optimization are presented here. Fig. 6 shows the results of a
V. R ESULTS building MPC controller using building optimization model-
The building model is implemented in MATLABr and I. The main objective in the building-side optimization is to
YALMIP [45] toolbox is used to provide a symbolic syntax minimize building electricity costs. Fig. 6 (a) demonstrates the
to formulate the problem and interfaces with the optimization temperature profile of the room, and Fig. 6 (b) shows load dis-
solver. Distribution system optimization model is developed in tribution of buildings connected to node #10 including lighting
GAMS [46]. The accuracy of the building model is validated load, appliances (computers, elevators, refrigerators, etc.) and
using actual measurements collected from an office building buildings’ HVAC load. Building load distribution is based on
at Michigan Tech [2]. The distribution grid model is validated simulation results carried out for Lakeshore Center building at
using the results in [11]. Michigan Tech using Carrier HAP software. Fig. 6(c) shows
This work considers the weather forecast and energy price load distribution on node #10. Fig. 6(d) illustrates the building
as the inputs required for the building and grid optimization electricity load and the maximum feasible load for node #10
models. Thus, electricity dynamic pricing available from Mid- which is obtained from distribution optimization model-I block
continent Independent System Operator (MISO) is used [47]. shown in Fig. 4. In Fig. 6(d), the importance of grid-wise
TABLE I
P ERFORMANCE OF FOUR DIFFERENT B2G OPTIMIZATION TECHNIQUES , COMPARED TO THE UNOPTIMIZED (RBC) CASE STUDY.
Optimization Monthly Bldg. Monthly Bldg. elec. Node #10 System (grid) IB2G Bldg. cost Bldg. energy
type elec. cost [$] consumption [kW h] load factor [-] load factor [-] index [-] saving* [%] saving* [%]
One day ahead prediction (including weather, dynamic pricing, comfort bounds, etc.)
Unoptimized (Base case) 935 368.1 0.41 0.72 10.50 0 0
Optimized (Bldg Unidirectional) 693 310.6 0.44 0.81 9.2 26 16
Optimized (Grid Unidirectional) 909 341.4 0.84 0.86 8.7 3 7
Optimized (Bidirectional) 698 310.1 0.46 0.84 8.8 25 17
Optimized (Using IB2G ) 707 305.2 0.67 0.84 6.9 24 17
* Percentage x−x0
saving is calculated by x0
× 100, where x0 is the energy consumption/cost for the unoptimized (base) case.
TABLE II
P ERFORMANCE OF TWO DIFFERENT B2G REAL - TIME OPTIMIZATION TECHNIQUES , COMPARED TO THE UNOPTIMIZED (RBC) CASE STUDY.
Optimization Monthly Bldg. Monthly Bldg. elec. Node #10 System (grid) IB2G Bldg. cost Bldg. energy
type elec. cost [$] consumption [kW h] load factor [-] load factor [-] index [-] saving* [%] saving* [%]
Real-time prediction (including weather, dynamic pricing, comfort bounds, etc.)
Optimized (Bidirectional) 752 308.9 0.47 0.81 9.8 20 16
Optimized (Using IB2G ) 758 308.6 0.59 0.83 7.9 19 16
* Percentage x−x0
saving is calculated by x0
× 100, where x0 is the energy consumption/cost for the unoptimized (base) case.
optimization for buildings is emphasized since the optimized

building loads violate the maximum allowable loads by the (a)
Temperature (o C)
grid. Note that in Fig. 6, the power and supply temperature 30

Comfort bound
Room temp.
peaks at unusual time is due to the pre-heating of the HVAC 25

Supply temp.
system when the electricity price is cheaper. The dynamic

20
pricing and result HVAC cost are shown in Fig. 7. The monthly
0 5 10 15 20 24
building electricity cost is listed in TABLE I. Compared to the (b)
Node‘s Buildings
150
RBC, building-side optimization results in 26% cost saving HVAC
load (kW)
Lighting
and 16% energy saving. 100
Appliances
Total
If the objective is to maximize load factor in the distribution 50
grid (i.e, the grid-side optimization), the nodal load factor is 0

significantly increased to 0.84 , while the building’s energy 0 5 10
(c)
15 20 24
Node load (kW)
cost increases. 150 Node total

Existing load
100 Bldgs' total load
C. Bidirectional Optimization
50
The proposed B2G methodology in Section III is applied 0
for the case study similar to that in subsection V-B. Building 0 5 10
(d)
15 20 24
150
optimization results are illustrated in Fig. 8 and summarized Buildings load
Load (kW)
Maximum feasible load for buildings

in TABLE I. The results show the B2G optimizer satisfies 100
the building comfort levels, while keeping the building load 50

under the maximum load allowed by the distribution grid. In
0
addition, compared to the unidirectional (demand side) opti- 0 5 10 15 20 24
Time (hour)
mization, the system level load factor is increased from 0.81
Fig. 6. Building-side optimization: (a) Control input and resulting temperature
to 0.84. IB2G index also depicts a significant improvement as profile for the building MPC controller, (b) Buildings’ load profile including
seen in TABLE I. Bidirectional optimizer offers 25% cost sav- HVAC load, lighting load and appliances load, (c) Buildings’ total load at
ing compared to the unoptimized case while energy cost saving node #10 and base load, (d) Buildings’ electricity load versus maximum
feasible load for node #10.
is only dropped by 1% compared to unidirectional (building
side). Thus, this case study demonstrates that with the grid
constraints, the cost of operation of buildings increases, but it
ensures a feasible operation of the grid. of bidirectional optimization using IB2G index are shown in
Fig. 9 and summarized in the TABLE I. As expected, the
D. Bidirectional Optimization Using IB2G Index new optimization approach leads to the best result in terms
IB2G index provides a way to control/formulate the im- of IB2G . The smaller value for IB2G , the better performance
portance of building benefits versus grid benefits. The results for the B2G system. Using IB2G in the building objective
(a)
Temperature (oC)
2 400
Dynamic pricing ($/MWh)

Electricity cost 30 Comfort bound
Dynamic pricing Room temp.
electricity cost ($)
Building’s HVAC
Supply temp.
1.5 300 25
20
1 200
0 5 10 15 20 24
(b)
0.5 100 150
Buildings load
Load (kW)
0 100
0
1 5 10 15 20 24
Time (hour)
50
Fig. 7. Building-side Optimization: dynamic pricing versus hourly cost for
the building. 0 5 10 15 20 24
Time (hour)
(a) Fig. 9. Bidirectional optimization based on IB2G index: (a) Control input
Temperature (oC)
and resulting temperature profile, (b) Building load versus maximum feasible
30 Comfort bound
Room temp.
load.
Supply temp.
25
20
E. Impact on Voltage Performance
Permissible voltage range, as specified in ANSI C84.1, is
0 5 10 15 20 24
one of the standards to maintain power quality in distribution
(b)
150 feeders. Fig. 10(a) shows the voltage at few selected nodes for
Buildings load
unidirectional optimization (building-side) which are violated
Load (kW)

100 in certain hours, and (b) illustrates that the optimization model
with IB2G index where voltages are maintained within the
50 limits.
0 5 10 15 20 24
(a)
Time (hour)
Voltage (p.u.)
Voltage lower bound Node #15 Node #17

0.98
Fig. 8. Bidirectional optimization: (a) Control input and room temperature
profile for the MPC controller, (b) Building load vs maximum feasible load. 0.96
0.94
function helps to improve building/nodal load factor and 1 5 10 15 20 24

(b)
consequently better system level load factor, while minimizing
Voltage (p.u.)
the building energy costs. As a result, IB2G is better in the 0.98 Voltage lower bound Node #15 Node #17
new bidirectional optimization, compared to the bidirectional

0.96
optimization with objective 8a. 0.5 and 0.54 are chosen for α
and β, respectively, to provide comparable weight to building 0.94
and grid. Selection of α and β can be another optimization
0 5 10 15 20 24
problem which is outside the scope of this work. In this case Time (hour)
study, the cost of electricity is decreased by 24% with respect
to the base case and both nodal and system load factors are Fig. 10. Voltage at selective nodes. Fig. (a) shows the voltage profile
significantly improved compared to the base case. of unidirectional optimization and Fig. (b) illustrates the voltage profile of
bidirectional optimization (using IB2G index).
If α in IB2G in the bidirectional optimization is chosen to
be zero, the bidirectional optimization becomes unidirectional
which mainly satisfies the grid benefits (Grid unidirectional
in TABLE I). As expected, this leads to the best load factor F. Hour-ahead vs Day-ahead Optimization
compared to other four cases in TABLE I. But, the buildings’ In the case studies in TABLE I, the prediction horizon
electricity cost is increased by 21% compared to the bidi- N is considered to be 24. It means that the optimization
rectional IB2G case. Overall, bidirectional optimization using algorithm has prediction of weather forecast and dynamic
IB2G provides an optimization platform in which an operator pricing for the next 24 hours. If the weather forecast or
can easily decide the desirable compromise for the benefits of prediction of electricity cost (dynamic pricing) is not available
customers and the grid. By using the proposed IB2G index as or not accurate for that period, the MPC algorithm cannot find
the optimization cost function, the temperature boundaries are the optimal solution for the next 24 hours. Here, we show two
met based on the ASHRAE standards. examples that the dynamic pricing and the weather forecast are
available for the next hour in TABLE II. Simulations for two Winter
(a)
proposed methods, bidirectional using (8a) and bidirectional 20 Fall
Temperature ( °C)
Spring
using IB2G , are carried out to demonstrate the effect of 10
prediction horizon on the cost saving and IB2G index. Results
0
show that B2G metrics deteriorate by reducing the prediction
horizon but the proposed algorithm still offers benefits for -10
both grid and buildings. Using the real-time B2G optimization,
-20
building electricity cost drops up to 20% and building load 1 5 10 15 20 24
Time (hour)
factor increases over 0.1 compared to the baseline unoptimized (b)
case. 400 MI
Dynamic pricing
IL
300 MN
($/MWh)
G. Impact of Dynamic Pricing and Seasonal Weather 200
The savings on electricity price and improvement on sys- 100

tem load factor greatly depend on various factors including
0
dynamic pricing, weather, flexibility of commercial build- 1 5 10 15 20 24
ing’s loads, building’s temperature set-points, available con- Time (hour)
trol equipment in the building and distribution grid, and

Fig. 11. (a) weather forecast of three sample days in Fall, Winter, and Spring
accuracy/availability of the forecasts. The impact of dynamic using measured data at Michigan Tech testbed and (b) Dynamic pricing for
pricing and weather variations on the customer’s cost savings Michigan, Illinois, and Minnesota obtained from [47].
and system load factor are also studied.
Fig. 11 (a) shows the weather condition for three sample
days in Fall, Winter, and Spring in our testbed. Fig. 11 (b)
illustrates three different dynamic energy prices at nodes in
Michigan, Illinois and Minnesota. Based on the weather and
energy price data, two sets of simulations are carried out.
Fig. 12 (a) shows the effect of weather on the building’s
electricity cost and system load factor for the three seasons
by keeping the dynamic energy price same (Michigan node).
Fig. 12 (b) demonstrates the effect of dynamic pricing on the
electricity cost saving and the system load factor improvement
using same weather forecast (Fall). During the Winter, 33%
saving in electricity cost and 8% improvement on system
load factor. However, as weather gets warmer, the cost sav-
ing decreases. For instance, in Spring, only 6% decrease in
electricity bill compared to RBC is observed. The reason for
less saving in Spring compared to Winter is that the cold
outside weather in Winter makes HVAC running more often
which provides more opportunities to save energy costs by
shifting the load. Less constraints (i.e., heating requirement) Fig. 12. (a) shows the weather effect on the building’s electricity bill and
on the HVAC controller in Spring/Fall compared to Winter system load factor, and (b) illustrates the effect of dynamic pricing on the
electricity bill and system load factor for the conditions shown in Fig. 11. An
provides more flexibility for load factor optimization. Thus RBC structure similar to that in TABLE I is used as a baseline to calculate
the percentage of increase in the B2G system’s LF is more saving percentages.
in Fall/Spring compared to that in the Winter. Finally, the
results in Fig. 12 (b) show that the proposed B2G optimization
provides cost saving and system LF increase for all the three computational challenges associated with the large scale non-
dynamic pricing profiles studied. convex optimization of distribution grid [49]. However, for
the size of the system considered in this paper, computational
complexity was not an issue. Simulation time for bidirectional
H. Computational Cost optimization discussed in this paper on an INTEL Core i5,
The problem formulation is hierarchical in nature, which 3.2 GHz CPU desktop computer is less than one minute for
allows each BEMS to solve its own building optimization building’s unidirectional optimization and around 5 minutes
model. The developed building optimization model in Section for optimization of both building and grid models. The entire
II is linear programming in nature, thus the computational hierarchical optimization problem takes maximum of two
tasks at the building level are not challenging for the demand iterations to solve. First, optimal load profiles at each interval
response applications discussed in this work. At the grid level, (i.e., each hour in this work) from the buildings are sent to grid
the nature of the problem is non-linear and LTCs and cap control center to determine feasibility of the grid operation. In
banks add integer variables in the model. This poses inherent case of infeasibility, maximum feasible load profiles are sent
to each building controllers. for the low energy prices, the impact of low energy price on
Computational challenges associated with large practical peak demand can be eliminated to a certain extent.
size grid can be reduced by using methods of convexifica-
tion [50], distributed approaches [14], and heuristic approaches VI. C ONCLUSION
[10]. The solution time desired for the proposed models This study develops a bidirectional B2G optimization frame-
depend on the B2G applications sought. For example, voltage work using an experimentally validated building thermal
regulation needs to be tackle in a few minutes, and load model and a mathematical model for the distribution grid. The
shifting in the order of hours [51]. For the type of B2G building optimization model is based on disturbance and heat
applications proposed in this work, solution time of 5 minutes transfer phenomena, and its model parameters are obtained
and a coarse time interval of one hour suffice. from actual measurements collected from an office building
at Michigan Technological University. The distribution grid
I. Benefits, Challenges, and Limitations model is comprised of modeling of typical distribution sys-
tem components including feeders, transformers, and control
This work provides a generic mathematical framework to equipment such as capacitor banks and transformer load tap
optimally coordinate building loads and grid level assets, changers. In the proposed bidirectional optimization model,
which is useful for near-term energy usage planning and/or the objective is to minimize energy cost for the demand side
near-real time dispatch of building loads. This opens up and to maximize load factor for the grid. To account for
opportunities to deploy multiple distributed building loads for conflicting interests of the BEMS and distribution operators in
grid level applications such as demand response, load fol- the bidirectional optimization, a novel B2G index is developed
lowing, and regulation services. The framework ensures that, based on building energy cost and nodal load factor. Based
in the demand dispatch process, the operational constraints on the provided case studies, it can be concluded that the
and interests of the grid level and customer level energy developed bidirectional optimization framework can reduce
management activities are honored; thus providing benefits to commercial buildings’ monthly electricity costs by 25% in
both the entities involved. Winter, compared to the unoptimized rule-based control of
The case studies clearly demonstrated the benefits of the the building loads, while improving the system load factor.
proposed framework to the building and grid operations. However, the savings obtained in energy price and improve-
However, the major challenges in large scale deployment of ment on system load factor greatly depend on various factors
B2G integration are: (1) infrastructural challenges including including energy price, flexibility of commercial building
interoperability of tools at building and customer levels, band- loads, customers’ preferences, available control equipment in a
width limitation, compatibility of system to handle different distribution grid, accuracy/availability of weather forecast and
data with different resolution and with different communica- dynamic pricing prediction, etc. Nevertheless, the developed
tion standards, and (2) mechanism barriers including lack of bidirectional optimization framework certainly offers benefits
financial models for costumer motivation, accurate predication to the customers and the utilities in B2G integration.
of weather and energy price, computational issues for grid op-
timization for real-time applications in practical-sized systems, ACKNOWLEDGMENT
and scale-ability of the control and optimization algorithms Special thanks to Mr. Gregory Kaurala from MTU Energy
[51]. Management Office for help in collecting the building exper-
Use of dynamic pricing (day ahead or real-time) at small imental data.
residential and commercial buildings has already begun [52],
[53]. With the widespread implementation of dynamic pricing R EFERENCES
for small customers, and with energy management systems
[1] DOE. (2014, Dec 10). Building Energy Data Book. [Online]. Available:
at customer’s premise, the power peak in distribution systems http://buildingsdatabook.eren.doe.gov
will shift to the hours with least expensive energy prices. Thus, [2] M. Maasoumy, M. Razmara, M. Shahbakhti, and A. S. Vincentelli,
the utility needs to devise demand limits for the customers for “Handling model uncertainty in model predictive control for energy
efficient buildings,” J. of Energy and Buildings, vol. 77, 2014.
the feasible grid operation. However, with the demand limits, [3] S. Privara, J. Cigler, Z. Váňa, F. Oldewurtel, C. Sagerschnig, and
energy savings of the customer will reduce. An appropriate E. Žáčeková, “Building modeling as a crucial part for building predictive
incentive/compensation mechanism need to take place [54], control,” J. of Energy and Buildings, vol. 56, pp. 8–22, 2013.
[4] J. Ma, S. J. Qin, B. Li, and T. Salsbury, “Economic Model Predictive
where utilities may compensate the customers for the reduc- Control for Building Energy Systems,” in Proceeding IEEE PES Inno-
tion in the savings. However, the compensation may not be vative Smart Grid Technologies, Detroit, MI, 2011.
necessary for all cases, since the case studies demonstrated [5] M. Pcolka, E. Zacekova, R. Robinett, S. Celikovsky, and M. Sebek,
“Economical Nonlinear Model Predictive Control for Building Climate
that the optimized energy costs of the buildings are reduced Control,” in Proceeding of American Control Conference (ACC), 2014.
compared to the RBC even with the grid constraints. Utilities [6] A. Brooks, E. Lu, D. Reicher, C. Spirakis, and B. Weihl, “Demand
need to devise subscription plans where the customers agree dispatch,” Power and Energy Magazine, vol. 8, no. 3, pp. 20–29, 2010.
[7] M. C. Bozchalui, S. A. Hashmi, H. Hassen, C. A. Cañizares, and
that the BEMSs honor the demand limits sent by the DSO and K. Bhattacharya, “Optimal Operation of Residential Energy Hubs in
penalties for noncompliance. The another approach to solve Smart Grids,” IEEE Transactions on Smart Grid, vol. 3, no. 4, pp. 1755–
this issue is the use of distribution location marginal price 1766, 2012.
[8] S. Shao, M. Pipattanasomporn, and S. Rahman, “Demand Response as a
(DLMP) [55]. If appropriate method for setting DLMP can be Load Shaping Tool in an Intelligent Grid With Electric Vehicles,” IEEE
designed, which considers the customer’s expected response Transactions on Smart Grid, vol. 2, no. 4, pp. 624–631, Dec. 2011.
[9] J. Taylor, A. Maitra, M. Alexander, D. Brooks, and M. Duvall, “Evalua- with Renewable Energy,” in Proceeding of IEEE Innovative Smart Grid
tions of Plug-in Electric Vehicle Distribution System Impacts,” in Proc. Technologies Conference Europe, 2014, pp. 1–6.
IEEE Power and Energy Society General Meeting, 2010. [33] B. Jin, P. Nuzzo, M. Maasoumy, Y. Zhou, and A. Sangiovanni-
[10] S. Paudyal, C. A. Cañizares, and K. Bhattacharya, “Optimal Operation Vincentelli, “A Contract-based Framework for Integrated Demand Re-
of Distribution Feeders in Smart Grids,” IEEE Transactions on Industrial sponse Management in Smart Grids,” in Proc. 2nd ACM International
Electronics, vol. 58, no. 10, pp. 4495–4503, Oct. 2011. Conference on Embedded Systems for Energy-Efficient Built Environ-
[11] S. K. Goswami and S. K. Basu, “A New Algorithm for the Reconfigura- ments, 2015, pp. 167–176.
tion of Distribution Feeders for Loss Minimization,” IEEE Transactions [34] J. Hagerman, G. Hernandez, A. Nicholls, and N. Foster, “Buildings-to-
on Power Delivery, vol. 7, no. 3, pp. 1484–1491, Jul. 1992. Grid Technical Opportunities: Introduction and Vision,” US Department
[12] Y. Deng, X. Ren, C. Zhao, and D. Zhao, “A Heuristic and Algorithmic of Energy, Energy Efficient and Renewable Energy (EERE), Tech. Rep.,
Combined Approach for Reactive Power Optimization with Time- 2014.
Varying Load Demand in Distribution Systems,” IEEE Transactions on [35] S. Bera, S. Misra, and J. Rodrigues, “Cloud computing applications for
Power Systems, vol. 17, no. 4, pp. 1068–1072, Oct. 2002. smart grid: A survey,” IEEE Transactions Parallel Distribution Systems,
[13] I. Roytelman, B. K. Wee, and R. L. Lugtu, “Volt/var control algorithm vol. 26, no. 5, pp. 1477–1494, May 2015.
for modern distribution management system,” IEEE Transactions Power [36] “VOLTTRON: An Intelligent Agent Platform for the Smart Grid.”
Systems, vol. 10, no. 3, pp. 1454–1460, 1995. [37] W. Khamphanchai, A. Saha, K. Rathinavel, M. Kuzlu, M. Pipattana-
[14] E. Dall’Anese, H. Zhu, and G. Giannakis, “Distributed optimal power somporn, S. Rahman, B. Akyol, and J. Haack, “Conceptual architecture
flow for smart microgrids,” IEEE Transactions on Smart Grid, vol. 4, of building energy management open source software (BEMOSS),” in
no. 3, pp. 1464–1475, 2013. Innovative Smart Grid Technologies Conference Europe, 2014, pp. 1–6.
[15] M. Liu, C. Cañizares, and W. Huang, “Reactive Power and Voltage [38] G. Ghatikar and R. Bienert, “Smart grid standards and systems interop-
Control in Distribution Systems With Limited Switching Operations,” erability: a precedent with OpenADR,” in Grid Interop Forum, 2011.
IEEE Transactions Power Syst., vol. 24, no. 2, pp. 889–899, 2009. [39] M. Razmara, G. Bharati, M. Shahbakhti, S. Paudyal, and R. D. Robi-
[16] M. Avci, M. Erkoc, and S. S. Asfour, “Residential HVAC load control nett III, “Bidirectional Optimal Operation of Smart Building-to-Grid
strategy in real-time electricity pricing environment,” in Proc. Ener- Systems,” in American Control Conference (ACC), 2015, pp. 288–293.
gytech, May 2012. [40] M. Maasoumy, M. Razmara, M. Shahbakhti, and A. Sangiovanni-
[17] D. T. Nguyen and L. B. Le, “Joint Optimization of Electric Vehicle and Vincentelli, “Selecting Building Predictive Control Based on Model
Home Energy Scheduling Considering User Comfort Preference,” IEEE Uncertainty,” in American Control Conference (ACC), 2014.
Transactions on Smart Grid, vol. 5, no. 1, pp. 188–199, Jan. 2014. [41] W. H. Kersting, Distribution System Modeling and Analysis. Chapters 3-
[18] Y. Yan, J. Zhou, Y. Lin, W. Yang, P. Wang, and G. Zhang, “Adaptive 6, Boca Raton, FL: CRC press, 2012.
optimal control model for building cooling and heating sources,” J. [42] E. Sortomme, M. M. Hindi, S. D. J. MacPherson, and S. S. Venkata,
Energy and Buildings, vol. 40, no. 8, pp. 1394–1401, 2008. “Coordinated charging of plug-in hybrid electric vehicles to minimize
[19] M. Maasoumy, A. Pinto, and A. Sangiovanni-Vincentelli, “Model-based distribution system losses,” IEEE Transactions on Smart Grid, vol. 2,
Hierarchical Optimal Control Design for HVAC Systems,” in ASME no. 1, pp. 198–205, Mar. 2011.
Dynamic Systems and Control Conference (DSCC), 2011, pp. 271–278. [43] R. F. Chang and C. N. Lu, “Feeder reconfiguration for load factor
[20] M. Razmara, M. Maasoumy, M. Shahbakhti, and R. D. Robinett III, improvement,” in Proc. Power Engineering Society Winter Meeting,
“Optimal exergy control of building HVAC system,” J. Applied Energy, vol. 2, 2002, pp. 980–984.
vol. 156, pp. 555–565, 2015. [44] A. Bokhari, A. Alkan, R. Dogan, M. Diaz-Aguilo, F. de Leon,
[21] X. Xu, S. Wang, and G. Huang, “Robust MPC for temperature control D. Czarkowski, Z. Zabar, L. Birenbaum, A. Noel, and R. Uosef, “Ex-
of air-conditioning systems concerning on constraints and multitype un- perimental determination of the zip coefficients for modern residential,
certainties,” J. Building Services Engineering Research and Technology, commercial, and industrial loads,” IEEE Transactions Power Delivery,
vol. 31, no. 1, pp. 39–55, 2010. vol. 29, no. 3, pp. 1372–1381, Jun. 2014.
[22] M. Razmara, M. Maasoumy, M. Shahbakhti, and R. D. Robinett III, [45] J. Löfberg, “YALMIP: A toolbox for modeling and optimization in
“Exergy-based Model Predictive Control for Building HVAC systems,” MATLAB,” in IEEE International Symposium on Computer Aided
in American Control Conference (ACC), 2015. Control Systems Design, 2004, pp. 284–289.
[23] G. Huang, S. Wang, and X. Xu, “A robust model predictive control [46] GAMS Development Corporation. (2014, Dec 10). General
strategy for improving the control performance of air-conditioning Algebric Modeling System, Release 24.2.1. [Online]. Available:
systems,” J. Energy Conversion and Management, vol. 50, no. 10, pp. http://www.gams.com
2650–2658, 2009. [47] MISO. (2014, Sep 3). Midcontinent Inde-
[24] F. Oldewurtel, A. Parisio, C. Jones, M. Morari, D. Gyalistras, M. Gw- pendent System Operator. [Online]. Available:
erder, V. Stauch, B. Lehmann, and K. Wirth, “Energy efficient building https://www.misoenergy.org/MarketsOperations/Prices/Pages/Prices.aspx
climate control using Stochastic Model Predictive Control and weather [48] ASHRAE, “Standard 62.1-2004,” Ventilation for Acceptable Indoor Air
predictions,” in Proceeding of American Control Conference (ACC), Quality, 2004.
June 2010, pp. 5100–5105. [49] S. Paudyal, C. Cañizares, and K. Bhattacharya, “Three-phase Distribu-
[25] Y. Ma, S. Vichik, and F. Borrelli, “Fast Stochastic MPC with Optimal tion OPF in Smart Grids: Optimality versus Computational Burden,” in
Risk Allocation Applied to Building Control Systems,” in IEEE 51st Proc. Innovative Smart Grid Technologies Europe, Dec. 2011, pp. 1–7.
Annual Conference on Decision and Control, Dec. 2012, pp. 7559–7564. [50] S. Low, “Convex Relaxation of Optimal Power Flow –Part I: Formu-
[26] P. D. Moroşan, R. Bourdais, D. Dumur, and J. Buisson, “Building lations and Equivalence,” IEEE Transactions on Control of Network
temperature regulation using a distributed model predictive control,” J. Systems, vol. 1, no. 1, pp. 15–27, Mar. 2014.
of Energy and Buildings, vol. 42, no. 9, pp. 1445–1452, 2010. [51] T. Kuruganti and M. Brambley, “Building-to-grid technical opportunity:
[27] Y. Ma, G. Anderson, and F. Borrelli, “A Distributed Predictive Control From the information and communications technology perspective,”
Approach to Building Temperature Regulation,” in Proceeding American DOE Energy Efficiency & Renewable Energy, Accessed on Feb 18, 2016.
Control Conference (ACC), 2011, pp. 2089–2094. [52] Ameren. Power Smart Pricing; A Smart Electricity Rate From Ameren
[28] M. Maasoumy and A. Sangiovanni-Vincentelli, “Buildings to Grid Illinois. [Online]. Available: https://www.powersmartpricing.org/
Integration: A Dynamic Contract Approach,” in Proc. IEEE/ACM In- [53] The COMED Residential Real-time Pricing Pro-
ternational Conference on Computer-Aided Design, 2015, pp. 473–478. gram Guide. Commonwealth Edison (ComEd). [Online].
[29] T. Gamauf, T. Leber, K. Pollhammer, and F. Kupzog, “A Generalized Available: https://www.comed.com/documents/customer-service/rates-
Load Management Gateway Coupling Smart Buildings to the Grid,” in pricing/real-time-pricing/rrtpprogramguide.pdf
Proc. AFRICON Conf., 2011. [54] Z. Wang, “Compensation Methods for Demand Response,” Master’s
[30] P. Palensky and D. Dietrich, “Demand Side Management: Demand thesis, University of Kentucky, 2015.
Response, Intelligent Energy Systems, and Smart Loads,” IEEE Trans- [55] R. Li, Q. Wu, and S. Oren, “Distribution locational marginal pricing for
actions on Industrial Informatics, vol. 7, no. 3, pp. 381–388, Aug. 2011. optimal electric vehicle charging management,” IEEE Transactions on
[31] X. Xue and W. Shengwei, “Interactive Building Load Management for Power Systems, vol. 29, no. 1, pp. 203–211, Jan. 2014.
Smart Grid,” in Proceeding of IEEE Power Engineering and Automation
Conference, 2012, pp. 1–5.
[32] A. Saha, M. Kuzlu, W. Khamphanchai, M. Pipattanasomporn, S. Rah-
man, O. Elma, U. S. Selamogullari, M. Uzunoglu, and B. Yagcitekin,
“A Home Energy Management Algorithm in a Smart House Integrated

Bilevel Optimization Framework For Smart Building-To-Grid Systems

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Bilevel Optimization Framework For Smart Building-To-Grid Systems

Încărcat de

Drepturi de autor:

Formate disponibile

This article has been accepted for publication in a future issue of this journal, but has not been

Bilevel Optimization Framework for

B UILDING sector in the United States accounts for

bidirectional information exchange cloud computing can be

Fig. 1. Summary of past studies on building, grid, and B2G optimization.

HVAC systems independently. However, there is little work

dTir 1 X h Tk − Tir i xk+1 = Axk + Buk + Ewk (5)

(a) output constraint on temperature of room and (8g) represents

20 should hold for all k = 0, 1, ..., N − 1, and N is the prediction

STD of error: 14.6 W

III. S OLUTION M ETHOD

grid power flow analysis is carried out to confirm the feasibility

optimization for buildings is emphasized since the optimized

grid. Note that in Fig. 6, the power and supply temperature 30

peaks at unusual time is due to the pre-heating of the HVAC 25

system when the electricity price is cheaper. The dynamic

grid (i.e, the grid-side optimization), the nodal load factor is 0

cost increases. 150 Node total

Maximum feasible load for buildings

the building comfort levels, while keeping the building load 50

Dynamic pricing ($/MWh)

Maximum feasible load for buildings

Voltage lower bound Node #15 Node #17

function helps to improve building/nodal load factor and 1 5 10 15 20 24

new bidirectional optimization, compared to the bidirectional

The savings on electricity price and improvement on sys- 100

trol equipment in the building and distribution grid, and

S-ar putea să vă placă și