=Paper=
{{Paper
|id=Vol-2785/paper12
|storemode=property
|title=Electricity Network Constraint Management using Individualised Demand Aware Price Policies
|pdfUrl=https://ceur-ws.org/Vol-2785/paper12.pdf
|volume=Vol-2785
|authors=I. Melatti,V. Alimguzhin,F. Mari,M. Prodanovic ,B. Hayes
|dblpUrl=https://dblp.org/rec/conf/overlay/MelattiAMPH20
}}
==Electricity Network Constraint Management using Individualised Demand Aware Price Policies ==
https://ceur-ws.org/Vol-2785/paper12.pdf
Proceedings of the
2nd Workshop on Artificial Intelligence and Formal Verification, Logics, Automata and Synthesis (OVERLAY),
September 25, 2020
Electricity Network Constraint Management using
Individualised Demand Aware Price Policies
I. Melatti 1 , V. Alimguzhin 1 , F. Mari 2 , M. Prodanovic 3 , and B. Hayes 4
1
Computer Science Dept., Sapienza University of Rome, Italy
2
Dept. of Movement, Human and Health Sciences, University of Rome “Foro Italico”, Italy
3
Electrical Systems Unit, IMDEA Energy Institute, Spain
4
Electrical and Electronic Engineering, University College Cork, Ireland
Abstract
Electric Distribution Network constraint management is employed by Distribu-
tion System Operators in order to keep inside desired safety bounds the aggregated
power demand at each network substation. In our context, such aggregated power
demand is due to residential users requiring electricity to the substation they are
connected to. This enables saving in substation maintenance and energy peak
production, as users typically tend to use little energy for most of the day, except
for demand peaks, especially during evenings. The main workhorse to obtain such
a goal is Demand Side Management, that is, trying to change the users demand in
order to meet aggregated demand safety bounds.
In this short paper, we introduce the problem and briefly review our recent
approach to perform Demand Side Management for Electric Distribution Network
constraint management, based on a network state estimator and a Model Predictive
Control scheme. We also show experimental results on large scenarios using a real
Electric Distribution Network in Denmark.
1 Introduction
An Electric Distribution Network (EDN) is composed of electrical substations, each servicing a number
of residential users. The typical users behaviour, as for electricity power demand, is to use little energy
for most of the day, and then request much electricity during the evening (demand peak). This increases
costs for Distribution System Operators (DSOs) managing the EDN, as it results in peak power plant
usage [37] and substation ageing [51]. In order to counteract such a problem, DSOs have to: i) detect,
for each substation, the desired safety power bounds for the aggregated power demand, resulting from
summing up the electricity power requested by all residential users connected to the same substation; and
ii) enforce such bounds. This activity is typical referred to as network constraint management.
The main tool DSOs use in order to minimise the aggregated power demand outside substation
desired bounds is Demand Side Management (DSM). Namely, DSOs try to modify the users power
demand by employing either i) Direct Load Control (DLC), i.e., directly acting on users appliances (see,
e.g., [39, 44, 20]) or ii) Autonomous Demand Response (DR), i.e., trying to economically incentivise users
to shift their demands (see Section 5 for a discussion of the literature). Namely, in the DR setting, each
Copyright © 2020 for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
76 I. Melatti, F. Mari
residential user receives a Time of Use (ToU) (i.e., prices may change at different hours of the day) and/or
Inclining Block Rate (IBR) (i.e., the per-kWh prices depend on the power demand itself) price policy.
Such price policy must be computed so as users aggregated demand is within substation desired bounds,
provided that most of the users follow their price policy. In order to follow the price policy, residential
users have to avoid energy consumption in hours where the price is high, and instead shift loads in hours
with low price. Finally, as for desired power bounds on substations, they are typically set up manually by
domain experts.
Unfortunately, this simple scheme has to face the peak rebound effect: as all users have the same price
policy, they tend to shift their demand in the same way, thus creating a new peak in the hours with low
tariff [16, 40, 56]. Furthermore, manually determining substation bounds is expensive and time consuming.
In this short paper, we introduce the problem and review our recently proposed approach [26, 14]
for the automatic computation of both desired bounds for substations (EDN Virtual Tomography, or
EVT, service) and of demand-aware individualised price policies for users (Demand Aware Price Policy, or
DAPP, service). Our approach is based on statistical methods such as Weighted Least Squares (WLS) [49]
for the EVT service and on the Model Predictive Control (MPC) methodology [43] relying on the GLPK
Mixed Integer Linear Programming (MILP) solver (gnu.org/software/glpk) for DAPP.
2 Problem requirements
An EDN is composed of several substations, where each substation serves a set of residential houses. We
suppose to be able, using the measurements taken from the home electricity mains (Advanced Metering
Infrastructure, AMI), to know each house power demand, with periodicity at least one hour. Our objective
is to reduce costs for the DSO, by limiting the demand drawn at some or all substations of the EDN at
times of peak demand. In fact, this reduces costs of buying energy from the market at times of peak
electricity price, and reduces overloading of network components during times of peak demand (thus
reducing substations aging), or during periods when the system is weakened due to line/transformer
maintenance or other outages [51, 37].
This must be done avoiding the peak rebound effect. A study carried out by a Danish DSO [46] in
2013 showed an example of such an effect. In this study, residential users of a Danish city were given a
ToU price policy, so as to test if the residential demand may be changed (actually, shifted) using only
economic incentives. Thus, such price policy gave electricity for free during nights (from 8PM to 6AM),
with a low price p during most part of the day (from 6AM to 5PM) and with an high price (more than
5p) during evening (from 5PM to 8PM, which was the usual electricity peak). As a result, the peak was
simply shifted: instead of occurring from 6PM to 7PM, it was from 8PM to 9PM, especially during Winter
months. These effects are undesirable, since DSOs objectives are to smooth the load profile, increase the
load factor (i.e., the ratio of average to maximum load), and reduce demand peaks.
3 Network constraint management through intelligent services
Our proposed architecture comprises the following two computational services [14, 26] (see Figure 1).
The EVT service uses available measurements of houses power demand (collected from Supervisory
Control And Data Acquisition, SCADA, and smart metering/AMI systems/electricity mains) and the
knowledge of the EDN topology to estimate, in real-time, the EDN state, using the WLS methodology [49].
Furthermore, EVT also carries out network analysis ahead of time, as described in [13, 11]. The results of
the state estimation and network analysis carried out by EVT is used to set operational constraints on
the EDN, by limiting the aggregated demand drawn at some or all substations within the EDN, especially
at times of peak demand.
The DAPP service takes as input both the output from EVT (i.e., the bounds on substations
aggregated demand) and again the power demand from houses of the EDN. Then, for each substation s in
the EDN, computes (by using a MILP solver and the MPC methodology) a set of individualised ToU and
IBR price policies. Such price policies are sent to users through the energy retailer (which may be the
DSO itself). Namely, the following holds. i) Each price policy defines a low tariff area and an high tariff
Electricity Network Constraint Management using Individualised Demand Aware Price Policies 77
EDN
Electrical Distr. + energy price policies
Substation
Network
measurements Load Factor Base Global Indiv.
EDN configuration
Network readings User flexibility Case Case Case
DSO Retailer
Distribution ! Winter Peak 0.7116 0.6176 0.8124
Operational price policies
Constraints
User flexibility
Summer Min 0.5927 0.5280 0.7038
Network Overall 0.3471 0.3084 0.4415
state
estimation Distribution price policies
Figure 2: Resulting Aggregated Load Factors
EVT DAPP
Services
Figure 1: Our overall architecture.
area, using the IBR methodology: if the power demand of the target house is within given power bounds,
then a low price of energy is applied; otherwise the high price is used. Thus, users are economically
incentivised to stay within the given power bounds. ii) Each price policy also follows the ToU scheme,
i.e., the bounds for low and high tariff area may change every hour. This allows not to compress the
overall demand, which would cause the DSO to lose money: a peak centred in a time slot is distributed
on the adjacent time slots. iii) The price policies are individualised, i.e., different houses will get possibly
different price policies. This avoids the peak rebound effect, as different houses are incentivised to shift
their demand to different time slots. iv) In order to have price policies which may be actually followed by
each user, we take into account each user flexibility, i.e., the capability of each house to shift its demand.
This also allows to design DAPP so as to output not-discriminatory policies: all residential users have the
same opportunities to always pay the low tariff.
4 Experimental results
We experimented our approach by using a case study from the European Commission project “SmartHG”
(smarthg.di.uniroma1.it) [29]. Such case study consists of a suburban/rural 10kV EDN with a weakly-
meshed structure, including 46 substations, each with 30 houses connected on average. We initially run
EVT to determine power bounds on aggregated power demand of all 46 substations in the EDN. Then,
we consider 3 different scenarios: the base case (data as actually recorded from house mains), the global
case (all houses are given the same “global” price policy) and the individualised case (all houses get the
individualised price policies output by DAPP).
As a result, the global price policy results in a rebound demand peak similar to that recorded in the
Danish study [46] described in Section 2. Instead, in the individual price policy case, the demand peaks
are much reduced due to the effect of the DAPP algorithm. This significantly improves the load factor, i.e.,
the ratio of average to maximum load. Figure 2 shows our results (the higher the load factor, the better).
This simulated increase in load factors due to the application of individualised price policies would have
clear benefits for the DSO. This would reduce the amount of energy to be purchased from the wholesale
market during expensive peak hours, and the flatter load profiles would result in less instances where
network is overloaded, potentially reducing network maintenance and upgrade costs and allowing deferral
of network investments. In fact, the global price policy results in heavier line loading values, whereas the
individualised price policy results in reduced line loading. Namely, the number of overloads drops from 18
(global case) to 2 (individualised case), whilst it was 3 in the base case.
5 Related work
DSM for EDNs is a well-studied topic on smart grids. It consists of two main methodologies: DLC and
DR. However, since here we focus on residential demand and DLC is mainly used to directly actuate large
industrial loads [44, 39] (though [20] shows a possible application to residential users), here we discuss
78 I. Melatti, F. Mari
DR [53, 45, 12, 1, 26, 6, 7]. As discussed above, the main drawback to DR schemas using global price
policies, i.e., in which all residential users get the same policy, is the peak rebound effect [16, 40, 56]. An
alternative strategy is to expose residential users to wholesale market prices (real-time pricing). However,
this causes demand to be shifted to hours with low electricity price, which can lead to a higher peak
electricity price and peak-to-average ratio during the low price time [56]. There is a significant challenge
in ensuring that such real-time prices do not cause physical or market instabilities [55], and it has been
shown that multiple, uncoordinated responses to frequently changing prices can cause increased volatility,
and potentially grid instability [17, 42].
Another problem is given by the fact that users may not follow (or too loosely follow) their price policy.
Several approaches have been proposed to overcome such drawback [19], such as the “aggregator” [38]
and “virtual power plant” concepts [41]. An alternative way is to compute the probability of failing in
keeping the aggregated demand within substation safety constraints, given a reasonable probability of
user deviations from their price policies, and show that such failure probability is low enough [36, 26],
using statistical and model checking based techniques [31, 8, 28, 25].
Finally, from a computational point of view, our approach (in particular, the DAPP service) is based
on several mature methodologies, such as MPC [43] as well as Artificial Intelligence and model checking
using (possibly mixed linear) constraint solving [10, 23, 24, 22, 35], which we successfully exploited in
very different areas, e.g., vulnerability of large networks to attacks [34] and in silico medicine, where
we performed simulation-based (e.g., [21]) synthesis and safety/efficacy assessment of pharmacological
treatments [47, 33, 27, 50], in an area (assisted reproduction) with low average success rates and many
factors not under full control [15].
6 Conclusions
In this short paper we introduced the problem of computing and enforcing constraints on an EDN.
Such constraints are expressed as bounds on the aggregated power demand resulting at each EDN
substation. For such a problem, we surveyed our recent approach [14, 26] which, using two interoperating
computational services, first computes safety power bounds for each network substation and then outputs
a set of individualised (i.e., each user gets a different price policy) ToU and IBR price policies for each
house in the EDN, so as to enforce the power bounds at each substation. Our experimental results, using
real-world scenarios taken from a Danish area, show that our approach is able to improve the EDN load
factor, while avoiding the peak rebound effect caused by global price policies (i.e., all users get the same
price policy).
As future work, we plan to design a third service, to be run at users premises, that automatically
follows a given price policy, thus relieving the residential users from such a task. This may be done either
by automatically planning usage of smart appliances (see, e.g., [3], also in a centralised way [54, 2]) of
by automatically driving home batteries (see, e.g., [4, 18], also in a centralised way [52, 48]). A second
area of further research is to use model checking techniques to formally prove that our approach is
correct [30, 32, 9].
Acknowledgements This work was partially supported by: Italian Ministry of University and Research
under grant “Dipartimenti di eccellenza 2018–2022” of the Department of Computer Science of Sapienza
University of Rome; EC FP7 project SmartHG (Energy Demand Aware Open Services for Smart Grid
Intelligent Automation, 317761); INdAM “GNCS Project 2019”.
References
[1] C. Adika and L. Wang. Demand-side bidding strategy for residential energy management in a smart
grid environment. IEEE Trans Smart Grid, 5(4), 2014.
[2] R. Atia and N. Yamada. Sizing and analysis of renewable energy and battery systems in residential
microgrids. IEEE Trans Smart Grid, 7(3), 2016.
Electricity Network Constraint Management using Individualised Demand Aware Price Policies 79
[3] A. Basit, et al. Efficient and autonomous energy management techniques for the future smart homes.
IEEE Trans Smart Grid, 8(2), 2017.
[4] C. Chau, et al. Cost minimizing online algorithms for energy storage management with worst-case
guarantee. IEEE Trans Smart Grid, 7(6), 2016.
[5] Q. Chen, et al. MILP, pseudo-boolean, and OMT solvers for optimal fault-tolerant placements of
relay nodes in mission critical wireless networks. Fundam Inform, 2020.
[6] T. Chiu, et al. Optimized day-ahead pricing with renewable energy demand-side management for
smart grids. IEEE IoT J, 4(2):374–383, 2017.
[7] W. Chiu, et al. Pareto optimal demand response based on energy costs and load factor in smart grid.
IEEE Trans Ind Inf, 16(3):1811–1822, 2020.
[8] A. B. de Oliveira Dantas, et al. A component-based framework for certification of components in a
cloud of HPC services. Sci Comp Progr, 191, 2020.
[9] P. Duan, et al. A systematic mapping study on the verification of cyber-physical systems. IEEE
Access, 6, 2018.
[10] G. Gottlob, et al. Conditional constraint satisfaction: Logical foundations and complexity. In
IJCAI 2007.
[11] B. Hayes and M. Prodanovic. State forecasting and operational planning for distribution network
energy management systems. IEEE Trans Smart Grid, 7(2), 2016.
[12] B. Hayes, et al. Optimal power flow for maximizing network benefits from demand-side management.
IEEE Trans Power Sys, 29(4), 2014.
[13] B. Hayes, et al. A closed-loop state estimation tool for MV network monitoring and operation. IEEE
Trans Smart Grid, 6(4), 2015.
[14] B. Hayes, et al. Residential demand management using individualised demand aware price policies.
IEEE Trans Smart Grid, 8(3), 2017.
[15] B. Leeners, et al. Associations between natural physiological and supraphysiological estradiol levels
and stress perception. Front Psycol, 10, 2019.
[16] Y. Li, et al. Automated residential demand response: Algorithmic implications of pricing models.
IEEE Trans Smart Grid, 3(4), 2012.
[17] D. Livengood and R. Larson. The energy box: Locally automated optimal control of residential
electricity usage. Service Science, 1(1), 2009.
[18] J. M. Lujano-Rojas, et al. Optimizing daily operation of battery energy storage systems under
real-time pricing schemes. IEEE Trans Smart Grid, 8(1), 2017.
[19] O. Lutz, et al. Dynamic tariff design for a robust smart grid concept: An analysis of global vs. local
incentives. In ISGT-Europe 2017.
[20] J. Ma, et al. Residential load scheduling in smart grid: A cost efficiency perspective. IEEE Trans
Smart Grid, 7(2), 2016.
[21] F. Maggioli, et al. SBML2Modelica: Integrating biochemical models within open-standard simulation
ecosystems. Bioinformatics, 36(7), 2020.
[22] T. Mancini. Now or Never: Negotiating efficiently with unknown or untrusted counterparts. Fundam
Inform, 149(1-2), 2016.
[23] T. Mancini, et al. Evaluating ASP and commercial solvers on the CSPLib. Constraints, 13(4), 2008.
[24] T. Mancini, et al. Combinatorial problem solving over relational databases: View synthesis through
constraint-based local search. In SAC 2012. ACM.
[25] T. Mancini, et al. Anytime system level verification via random exhaustive hardware in the loop
simulation. In DSD 2014. IEEE.
[26] T. Mancini, et al. Demand-aware price policy synthesis and verification services for smart grids. In
SmartGridComm 2014.
80 I. Melatti, F. Mari
[27] T. Mancini, et al. Computing biological model parameters by parallel statistical model checking. In
IWBBIO 2015, LNCS 9044. Springer.
[28] T. Mancini, et al. SyLVaaS: System level formal verification as a service. In PDP 2015. IEEE.
[29] T. Mancini, et al. User flexibility aware price policy synthesis for smart grids. In DSD 2015. IEEE.
[30] T. Mancini, et al. Anytime system level verification via parallel random exhaustive hardware in the
loop simulation. Microprocessors and Microsystems, 41, 2016.
[31] T. Mancini, et al. SyLVaaS: System level formal verification as a service. Fundam Inform, 1–2, 2016.
[32] T. Mancini, et al. On minimising the maximum expected verification time. Inf Proc Lett, 122, 2017.
[33] T. Mancini, et al. Computing personalised treatments through in silico clinical trials. A case study
on downregulation in assisted reproduction. In RCRA 2018, CEUR W.P. 2271. CEUR.
[34] T. Mancini, et al. An efficient algorithm for network vulnerability analysis under malicious attacks.
In ISMIS 2018. Springer.
[35] T. Mancini, et al. Optimal fault-tolerant placement of relay nodes in a mission critical wireless
network. In RCRA 2018, CEUR W.P. 2271. CEUR.
[36] T. Mancini, et al. Parallel statistical model checking for safety verification in smart grids. In
SmartGridComm 2018. IEEE, 2018.
[37] G. Masters, et al. Renewable and Efficient Electric Power Systems. Wiley, 2004.
[38] J. Medina, et al. Demand response and distribution grid operations: Opportunities and challenges.
IEEE Trans Smart Grid, 1, 2010.
[39] National Grid. URL www2.nationalgrid.com/uk/services/balancing-services/.
[40] Y. Ozturk, et al. An intelligent home energy management system to improve demand response. IEEE
Trans Smart Grid, 4(2), 2013.
[41] D. Pudjianto, et al. Virtual power plant and system integration of distributed energy resources. Ren
Power Gener, IET, 1, 2007.
[42] A. Rajabi, et al. Aggregation of small loads for demand response programs — implementation and
challenges: A review. In EEEIC/ICPS Eur 2017.
[43] J. Rawlings, et al. Model Predictive Control: Theory, Computation, and Design. Nob Hill , 2017.
[44] Red Electr ES www.ree.es/en/activities/operation-of-the-electricity-system/interruptibility-service.
[45] P. Samadi, et al. Tackling the load uncertainty challenges for energy consumption scheduling in smart
grid. IEEE Trans Smart Grid, 4(2), 2013.
[46] Win with new electricity habits ipaper.ipapercms.dk/SeasNVE/Winwithnewelectricityhabits/.
[47] S. Sinisi, et al. Optimal personalised treatment computation through in silico clinical trials on patient
digital twins. Fundam Inform, 2020.
[48] A. Soroudi, et al. Optimal dr and ess scheduling for distribution losses payments minimization under
electricity price uncertainty. IEEE Trans Smart Grid, 7(1), 2016.
[49] T. Strutz. Data Fitting and Uncertainty. 2010.
[50] E. Tronci, et al. Patient-specific models from inter-patient biological models and clinical records. In
FMCAD 2014. IEEE.
[51] M. Uddin, et al. A review on peak load shaving strategies. Ren Sust Energy Rev, 82, 2018.
[52] R. Wang, et al. A self-interested distributed economic model predictive control approach to battery
energy storage networks. J Proc Contr, 73, 2019.
[53] H. Yan, et al. Future evolution of automated demand response system in smart grid for low-carbon
economy. J Modern Power Sys Clean Energy, 3:72–81, 2015.
[54] E. Yao, et al. Residential demand side management under high penetration of rooftop photovoltaic
units. IEEE Trans Smart Grid, 7(3), 2016.
[55] X. Zhang, et al. Energy Solutions to Combat Global Warming. LN in Energy. Springer, 2016.
[56] Z. Zhao, et al. An optimal power scheduling method for demand response in home energy management
system. IEEE Tran Smart Grid, 4(3), 2013.