Contingency Analysis to Evaluate the Robustness in Large-Scale Smart Grids: Based on Information Security Objectives and Frequency Stability

Abstract

Smart grids (SGs) are attracting attention as high value-added platforms, which mass-produce new business models through real-time information sharing. However, the open-ended information structure of SGs increases the risk of exposure to cyberattacks through the creation of multiple communication access points. In a power system where a real-time balance of supply and demand is essential, cyberattacks result in cascading failures leading to power outages. Therefore, this paper proposes a method to evaluate the robustness of large-scale SGs against cybersecurity disturbances. The proposed evaluation method established a hierarchy quantification technique considering the structural characteristics of SGs. With respect to the cyber hierarchy, relevant standards (NERC CIP, NIST FIPS) were applied to classify the grades of information security risk. In the case of physical hierarchy, the power system was calculated by using optimal power flow and analyzed the frequency stability. This study was aimed at identifying the vulnerabilities in the physical topology aspect of intelligent power systems due to cybersecurity disturbances.

1. Introduction

A smart grid (SG) is a representative convergence technology that provides various services and high-value through the advancement of power infrastructure. However, due to the physical characteristics of electricity (maintaining a balance of real-time supply and demand, etc.), this industry faces complex and diverse challenges compared with other convergence industries. A typical example is the uncertainty of a stable power supply due to cyber terrorism in the power system. On 14 August 2003, a massive blackout in nine areas in the United States and Canada caused more than $6 billion in monetary losses [1]. Although this incident is not a direct case of cyber terrorism in the power system, it indicates that control system malfunctions in the power industry can lead to national damage. In the United States and Europe, information and communications technology (ICT) has been used owing to the vastness of the geographic area and the advancement of the power market and there have been various cyber-attacks [2,3,4]. Moreover, related international conferences (Black Hat) have also proven the technical possibility of cybersecurity threats to SGs [5].

Recently, with the emergence of security issues due to the active deployment of SGs, various studies have been conducted on the optimized security solutions of SGs. The Illinois Institute of Technology has conducted a study on a contingency-based power system reliability evaluation algorithm for the malfunction of WAMS [6]. In addition, the Schweitzer Engineering Lab has analyzed the CPS modeling-based vulnerabilities of the power infrastructure [7]. Furthermore, SG security studies have been reported from various perspectives, such as the analysis of the impact of cyber disturbances on the power market [8,9,10]. In recent years, related studies have been actively conducted on SG security reflecting cyber disturbances for the power-system-analysis techniques such as optimal power-flow-calculation normalization according to cyber-attacks [11,12].

In this study, the organic relationship of the SG infrastructure was analyzed, and a robustness-evaluation technique was proposed in terms of the physical topology based on the contingency analysis of cyber disturbances. To this end, Section 2 presents the analysis of the structural characteristics and infrastructure elements of the SG, and Section 3 introduces a physical robustness-evaluation method for a large-scale SG with respect to cyber disturbances. Particularly, for the evaluation, the cyber and physical hierarchies are quantified to reflect the structural characteristics of SGs. Section 4 validates the proposed evaluation method through a case study for the modified IEEE RTS 39-bus.

2. Structural Characteristics of SG

An SG is a convergence of power systems and information systems. Hence, the relationship between different fields needs to be analyzed. In this study, the correlation between the physical hierarchy (power system) and cyber hierarchy (information system) is analyzed considering a cybersecurity disturbance based on the power system in Korea. The power system is organically connected to the system of power generation, transmission, and distribution, centered on the energy management system (EMS) to maintain a real-time balance of power supply and demand. The EMS is a core system for the optimal operation of the power system and various optimization programs are installed to control the overall process from the power generation plan to subsystem analysis. The communication system of the EMS is based on IEC 61850 and 60870 and it transmits and receives data to/from power generators exceeding 20 MW and substations of 345/765 kV through a dedicated remote terminal unit (RTU). Substations below 154 kV acquire data through the regional control center (RCC) [13].

The power generation system maintains the specified frequency (60 ± 0.2 Hz in Korea) by adjusting the output reference point for each generator through the automatic generation control (AGC) function of the EMS as a synchronous power generation method [14]. The control system of a power generator over 20 MW periodically measures and controls approximately 3000 to 50,000 data elements on generator components (boiler, turbine, condenser, etc.) through the distributed control system (DCS) [15].

The transmission system monitors and controls various types of information on transmission and substation facilities through supervisory control and data acquisition (SCADA) and communicates based on IEC 61850, in connection with wide-area monitoring and control (WAMAC). WAMAC minimizes the spread of failures by adjusting the power distribution with an interruptible load shedding (ILS) in case of sudden disturbance [16].

These structural characteristics of SGs and the real-time two-way information-sharing system increase the information access points and, thus, these systems become exposed to cybersecurity threats. Therefore, the analysis of the correlation between hierarchies according to the occurrence of disturbances in an SG is as follows.

Figure 1 shows the structural and functional aspects of the SG, categorized into physical, cyber, and applied hierarchies. The figure also represents the correlation of each hierarchy based on the state change (steady, transient, and collapse) of the power system. Abnormal conditions in the cyber and application layers are directly connected to system malfunction, disabling the functions of each hierarchy, and a failure of the communication system is likely to cause a wide-area power outage.

3. Evaluation Methodology on Risk Impact of Cybersecurity Threats in Large-Scale SGs

3.1. Definition of Risk Impact of Cybersecurity Threats

NIST and NERC have announced strategies and technical standards for securing the cybersecurity of integrated power and information systems for the reliable operation of SGs [17,18,19,20]. In particular, NERC stipulated bulk electric system (BES) operators to evaluate the effect of the power system periodically due to cyber disturbances [19]. This study defines the concept of the risk impact of cybersecurity threats specialized in SGs based on related previous studies.

First, the power system at the physical hierarchy adheres to the reliability, adequacy, and security evaluation to secure soundness [21]. Reliability encompasses adequacy, security, and the ability to provide quality power to consumers via normal operation of the related facilities according to the power system regulations [21,22]. Security is the ability of the electric system to withstand sudden disturbances, such as electric short circuits or unanticipated loss of system elements. For this, the stability of frequency, voltage, etc., due to contingency is analyzed [21,22].

Next, securing cybersecurity at the cyber hierarchy refers to a series of technologies that satisfy the three elements of security objectives (confidentiality, integrity, and availability) for the information and communication infrastructure [17]. Confidentiality prevents information from being disclosed to unauthorized devices or users. Integrity ensures that information has not been altered by unauthorized actions during data processing. Availability only allows timely access and the use of data and information systems to authorized users.

Accordingly, the risk impact of cybersecurity threats in SGs is the integration of the concepts of power system security and cybersecurity and is defined as identifying vulnerable points from the perspective of physical topology due to cyber contingency in a large-scale intelligent power system.

3.2. Evaluation of Risk Impact of Cybersecurity Threats

In this study, the evaluation of the robustness of cybersecurity threats was established in six steps, as shown in Figure 2, focusing on quantitatively deriving vulnerable points in terms of physical topology.

Step 1 is the basic work to identify the evaluation targets of risk impact. It defines the components of the physical hierarchy (power system) and cyber hierarchy (information system).

Step 2 is a quantification technique from the aspect of cyber hierarchy, and the potential damage size based on the three elements of security objectives (confidentiality, integrity, availability) is determined to estimate the information security risk. The security objectives were applied to the FIPS 199 of the NIST and CIP technical standards of the NERC [19,23]. FIPS 199 defines the potential damage size based on the three elements of security objectives, as shown in

Table 1. The potential impact criteria for security objectives [17].

Security Objective	Potential Impact
	Low (L)	Medium (M)	High (H)
Confidentiality	The unauthorized disclosure of information could be expected to have a limited adverse effect on organizational operations, assets, or individuals.	The unauthorized disclosure of information could be expected to have a serious adverse effect on organizational operations, assets, or individuals.	The unauthorized disclosure of information could be expected to have a severe or catastrophic adverse effect on organizational operations, assets, or individuals.
Integrity	The unauthorized modification or destruction of information could be expected to have a limited adverse effect on organizational operations, assets, or individuals.	The unauthorized modification or destruction of information could be expected to have a serious adverse effect on organizational operations, assets, or individuals.	The unauthorized modification or destruction of information could be expected to have a severe or catastrophic adverse effect on organizational operations, assets, or individuals.
Availability	The disruption of access to, use of information, or an information system could be expected to have a limited adverse effect on organizational operations, assets, or individuals.	The disruption of access to, use of information, or an information system could be expected to have a serious adverse effect on organizational operations, assets, or individuals.	The disruption of access to, use of information, or an information system could be expected to have a severe or catastrophic adverse effect on organizational operations, assets, or individuals.

. CIP-002-5.1a (critical cyber asset categorization) is a standard for estimating the values for the power facility and classifies the values grade of each component of the SG. In other words, for the information security risk, the values of the components of the SG are calculated, and then the potential damage size is determined with respect to the malfunction of the corresponding equipment. Eleven security threat elements were considered for the SG infrastructure [24]; however, in the case of the occurrence of any one of these security threats, it is simplified as a malfunction of the corresponding equipment. Here, the simplification is identical in the contingency analysis of the power system (step 4), and accordingly, the robustness-evaluation result of SG comprising two hierarchies (information and physical hierarchies) is finally derived.

Step 3 estimates the optimal state of the power system by using the interior point method-based optimal power flow (IPM-OPF) to analyze the physical topology. The objective function of the IPM-OPF is to minimize the power generation costs and the AC power flow is considered as a constraint [25,26].

Objective function

$$\begin{gathered} \mathrm{Min}. \mathrm{F}={\displaystyle \sum }_{\mathrm{i}\in \mathrm{I}}{\displaystyle \sum }_{\mathrm{m}\in {\mathrm{M}}_{\mathrm{i}}}{\mathrm{f}}_{\mathrm{im}}, \forall \mathrm{m}\in {\mathrm{M}}_{\mathrm{i}} \\ {\mathrm{f}}_{\mathrm{im}}={\mathsf{\alpha }}_{\mathrm{im}}+{\mathsf{\beta }}_{\mathrm{im}}\cdot {\mathrm{PG}}_{\mathrm{im}}+{\mathsf{\gamma }}_{\mathrm{im}}\cdot {\mathrm{PG}}_{\mathrm{im}}^2 \end{gathered}$$

(1)

Constraints

$${\displaystyle \sum }_{\mathrm{m}\in {\mathrm{M}}_{\mathrm{i}}}{\mathrm{PG}}_{\mathrm{im}}+{\displaystyle \sum }_{\mathrm{j}\in \mathrm{I}}{\mathrm{PF}}_{\mathrm{ij}\hspace{0.33em}}={\mathrm{PL}}_{\mathrm{i}},\hspace{0.33em}\forall \mathrm{i}\in \mathrm{I}$$

(2)

$${\displaystyle \sum }_{\mathrm{m}\in {\mathrm{M}}_{\mathrm{i}}}{\mathrm{QG}}_{\mathrm{im}}+{\displaystyle \sum }_{\mathrm{j}\in \mathrm{I}}{\mathrm{QF}}_{\mathrm{ij}}={\mathrm{QL}}_{\mathrm{i}},\hspace{0.33em}\forall \mathrm{i}\in \mathrm{I}$$

(3)

$${\mathrm{PF}}_{\mathrm{ij}}={\mathrm{V}}_{\mathrm{i}}{\mathrm{V}}_{\mathrm{j}}\left\{-{\mathrm{G}}_{\mathrm{l}}\cos \left({\mathsf{\delta }}_{\mathrm{i}}-{\mathsf{\delta }}_{\mathrm{j}}\right)+{\mathrm{B}}_{\mathrm{l}}\sin \left({\mathsf{\delta }}_{\mathrm{i}}-{\mathsf{\delta }}_{\mathrm{j}}\right)\right\}+{\left({\mathrm{V}}_{\mathrm{i}}\right)}^2{\mathrm{G}}_{\mathrm{l}},\hspace{0.33em}\forall \mathrm{l}$$

(4)

$${\mathrm{QF}}_{\mathrm{ij}}=-{\mathrm{V}}_{\mathrm{i}}{\mathrm{V}}_{\mathrm{j}}\left\{{\mathrm{G}}_{\mathrm{l}}\sin \left({\mathsf{\delta }}_{\mathrm{i}}-{\mathsf{\delta }}_{\mathrm{j}}\right)+{\mathrm{B}}_{\mathrm{l}}\cos \left({\mathsf{\delta }}_{\mathrm{i}}-{\mathsf{\delta }}_{\mathrm{j}}\right)\right\}+{\left({\mathrm{V}}_{\mathrm{i}}\right)}^2\left({\mathrm{B}}_{\mathrm{l}}-{\mathrm{B}}_{\mathrm{c} \mathrm{a} \mathrm{p}}/2\right),\hspace{0.33em}\forall \mathrm{l}$$

(5)

$${\mathrm{PF}}_{\mathrm{ij}}\leqq {\mathrm{TP}}_{\mathrm{l}},\hspace{0.33em}\forall \mathrm{l}$$

(6)

$${\mathrm{QF}}_{\mathrm{ij}}\leqq {\mathrm{TQ}}_{\mathrm{l}},\hspace{0.33em}\forall \mathrm{l}$$

(7)

$${\mathrm{PG}}_{\mathrm{im}}^{\min }\leqq {\mathrm{PG}}_{\mathrm{im}}\leqq {\mathrm{PG}}_{\mathrm{im}}^{\max },\hspace{0.33em}\forall \mathrm{m}\in {\mathrm{M}}_{\mathrm{i}}$$

(8)

$${\mathrm{QG}}_{\mathrm{im}}^{\min }\leqq {\mathrm{QG}}_{\mathrm{im}}\leqq {\mathrm{QG}}_{\mathrm{im}}^{\max },\hspace{0.33em}\forall \mathrm{m}\in {\mathrm{M}}_{\mathrm{i}}$$

(9)

$${\mathrm{V}}_{\mathrm{i}}^{\min }\leqq {\mathrm{V}}_{\mathrm{i}}\leqq {\mathrm{V}}_{\mathrm{i}}^{\max },\hspace{0.33em}\forall \mathrm{i}\in \mathrm{I}$$

(10)

Step 4 analyzes the physical state of the power system in response to a contingency. The frequency of the power system and the voltage of each bus are profiled based on the malfunction of power facilities, such as generators and transformers, owing to cyber disturbance.

Step 5 evaluates the robustness and vulnerability of SGs by indexing the quantification results for each hierarchy. Information security risk indexing applies the result of Step 2. Frequency stability indexing rates the nadir and steady-state frequencies in the primary frequency response of the power system, as shown in Figure 3 [27]. The rating criteria (

Table 2. Risk criteria of frequency stability.

Time	Risk	Frequency Response
60 s (Primary response)	L (caution)	Nadir: 59.7 Hz or above (able to recover 59.8 within 1 min)
	M (warning)	Nadir: 59.3 Hz or above (able to recover 59.5 within 1 min)
	H (severe)	Nadir: below 59.3 Hz (unable to recover 59.5 within 1 min)

) are defined based on the electricity market regulations of Korea [14].

Step 6 establishes a security strategy for the result of step 5 (vulnerability analysis) from the perspective of the national SG operator. However, as security strategies are a separate research field, in which technical and economic matters must be considered, the usual level of physical security countermeasures are presented according to the simulation results in this study.

4. Case Study

4.1. Simulation System Modeling and Prerequisites

This study performed a simulation by partially modifying the IEEE RTS 39-bus [28] to verify the evaluation methodology on the risk impact of cybersecurity threats in large-scale SGs. As a modification, the generators (10 units) of the simulation system were modeled as a 560 MW thermal power generator with the same mechanical characteristics (

Table 3. Cost coefficient and model of generator.

Cost Coefficient (a + bX + cX2)			Generator Model (PSS/E Standard)
a	b	c	Generator	Exciter	Governor
103.82	2.1	0.001	GENROU	IEEET1	IEEEG1

) [29]. The load of the simulation system (modified IEEE RTS 39-bus) is summarized in Appendix A (

Table A1. Load data of modified IEEE RTS 39-bus.

Bus No.	Load		Bus No.	Load
	Active Power MW	Reactive Power Mvar		Active Power MW	Reactive Power Mvar
Bus 3	300	2.4	Bus 23	200	84.6
Bus 4	400	184	Bus 24	300	−92.2
Bus 7	100	84	Bus 25	200	47.2
Bus 8	400	153	Bus 26	100	17
Bus 12	10	88	Bus 27	300	75.5
Bus 15	300	153	Bus 28	200	27.6
Bus 16	300	32.3	Bus 29	300	26.9
Bus 18	100	30	Bus 31	10	4.6
Bus 20	400	103	Bus 39	500	250
Bus 21	200	115

). The simulation system for the case study is shown in Figure 4. For the quantification of the power system, this study utilizes the Fortran-based IPM-OPF computation model and the power system analysis tool PSS/E.

① The main premise is as follows. In the simulation system, the SG components are limited to EMS and DCS, as shown in

Table 4. Definition of a generator in the simulated system.

Bus No.	Generator No.	Control System
Bus30	G1	DCS1
Bus31	G2	DCS2
Bus32	G3	DCS3
Bus33	G4	DCS4
Bus34	G5	DCS5
Bus35	G6	DCS6
Bus36	G7	DCS7
Bus37	G8	DCS8
Bus38	G9	DCS9
Bus39	G10	DCS10

.

② The SG security standard (FIPS 119, SGSF-121-1-1) and the reliability standard (NERC’s CIP-002) are applied for threat identification and information security risk calculation for each SG component.

③ A contingency is an N-1 level generator trip in the case of a cyber disturbance.

④ The reference frequency of the power system is the center of inertia (COI) frequency [30].

$${\mathrm{F}}_{\mathrm{COI}}= \frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{H}}_{\mathrm{i}}·{\mathrm{f}}_{\mathrm{i}}}{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{H}}_{\mathrm{i}}}$$

(11)

⑤ The frequency stability assessment is shown in Table 2.

4.2. Simulation Results and Analysis

The information security risk by the SG component is listed in

Table 5. Information security risk by SG component.

Information Security	EMS	DCS
Availability	H	M
Integrity	H	M
Confidentiality	H	M

by applying prerequisites ① and ②. Table 5 shows the result of applying the security risk (Table 1) based on NERC’s CIP-002 (Influence on the power system by the capacity of power facilities) after identifying the security threats defined in the SG security standard (SGSF-121-1) for SG components.

The optimal power generation of the simulation system and the frequency response characteristics due to the contingency analysis are listed in

Table 6. IPM-OPF based optimal power generation and frequency response for contingency.

Gen.	Optimal Power Generation MW	COI Frequency Hz	Risk
G1	467.69	Nadir frequency: 59.22667 Steady-state frequency: 59.78536	H
G2	469.57	Nadir frequency: 59.35164 Steady-state frequency: 59.80185	M
G3	466.87	Nadir frequency: 59.31843 Steady-state frequency: 59.79653	M
G4	457.01	Nadir frequency: 59.25537 Steady-state frequency: 59.79105	H
G5	455.54	Nadir frequency: 59.27007 Steady-state frequency: 59.79313	H
G6	463.25	Nadir frequency: 59.25750 Steady-state frequency: 59.79028	H
G7	460.69	Nadir frequency: 59.21126 Steady-state frequency: 59.78463	H
G8	463.58	Nadir frequency: 59.19435 Steady-state frequency: 59.78056	H
G9	470.43	Nadir frequency: 59.31809 Steady-state frequency: 59.79630	M
G10	470.47	Nadir frequency: 59.37225 Steady-state frequency: 59.80436	M

. The optimal power generation for each generator was calculated using the Fortran-based IPM-OPF computation model. The optimal power flow result is covered in the Appendix A (

Table A2. Optimal power flow result.

Bus i	Bus j	Capacity MW	Power Flow MW	Bus i	Bus j	Capacity MW	Power Flow MW	Bus i	Bus j	Capacity MW	Power Flow MW
1	2	600	−65.66	9	39	900	−35.89	19	33	900	−455.63
1	39	1000	65.66	10	11	600	204.39	20	34	900	−453.78
2	3	500	406.53	10	13	600	262.48	21	22	900	−445.55
2	25	500	−4.71	10	32	900	−466.87	22	23	600	16.15
2	30	900	−467.69	11	12	500	−7.86	22	35	900	−463.25
3	4	500	40.15	12	13	500	−2.14	23	24	600	275.85
3	18	500	64.32	13	14	600	260.03	23	36	900	−459.71
4	5	600	−189.86	14	15	600	89.07	25	26	600	257.58
4	14	500	−170.12	15	16	600	−211.07	25	37	900	−462.37
5	6	1200	−388.48	16	17	600	213.25	26	27	600	123.92
5	8	900	198.34	16	19	600	−505.52	26	28	600	40.12
6	7	900	266.81	16	21	600	−245.09	26	29	600	−8.45
6	11	480	−196.04	16	24	600	25.73	28	29	600	−159.95
6	31	1800	−459.57	17	18	600	35.74	29	38	1200	−468.75
7	8	900	166.31	17	27	600	176.73
8	9	900	−35.86	19	20	900	−53.75

). For the contingency analysis, the frequency stability for each tripped generator (N-1) was evaluated according to the prerequisites ③, ④, and ⑤. In the case of the generators G1, G4, G5, G6, G7, and G8 failing, the nadir frequency of the simulation system is less than 59.3 Hz and the frequency stability is grade H severity.

The evaluation results of robustness for the simulation system are shown in

Table 7. Result on evaluating risk impact.

SG Component	Risk of Information Security			Frequency Stability
	Availability	Integrity	Confidentiality
EMS	H	H	H	System collapse
DCS1	M	M	M	H
DCS2	M	M	M	M
DCS3	M	M	M	M
DCS4	M	M	M	H
DCS5	M	M	M	H
DCS6	M	M	M	H
DCS7	M	M	M	H
DCS8	M	M	M	H
DCS9	M	M	M	M
DCS10	M	M	M	M

. In the tables, all DCSs except those of the EMS were calculated to be an identical grade of M for the information security risk assessment of the cyber hierarchy. This is because of the capacity-oriented potential impact assessment of the power facility. However, when the results of the frequency stability, as given in Table 6 are applied, the evaluation shows that generators G2, G3, G9, and G10 correspond to grade M, and the other generators (G1, G4, G5, G6, G7, and G8) correspond to grade H. In addition, when a generator of grade H is tripped, the imbalance in the supply and demand is larger than that in the case of grade M generators, as shown in Figure 5. The supply and demand balance graphs of the power system for the remaining generator trips are included in the Appendix A. Through this case study, we confirmed that the impact on the power system differs depending on the geographical location of the generator and transmission line configuration, even for the contingency of the generator with the same specification.

5. Conclusions

The power industry is discussing ways to strengthen cybersecurity as a global agenda. Accordingly, the SG infrastructure was analyzed in this study, and a method to evaluate the robustness of large-scale SGs was proposed, in terms of physical topology by cyber disturbances.

The evaluation technique for the risk impact of cybersecurity threats proposed in this paper focused on analyzing the correlation between the cyber hierarchy (information system) and the physical hierarchy (power system) and deriving the vulnerability from the perspective of physical topology through the quantification of each hierarchy. Regarding the quantification methodology for each hierarchy, a potential impact assessment based on three elements of information security, optimal state estimation based on optimal power flow, and frequency stability analysis were used. Additionally, the objectivity and effectiveness of the study results are improved by applying technical standards (NERC CIP, FIPS) to the evaluation criteria and conducting case studies based on the Modified IEEE RTS 39-bus. This study aims to present a basis for optimal decision-making when establishing security countermeasures from the point of view of the national SG operator. A follow-up study requires specific correlation analysis of various security threats and SG infrastructure as well as case studies based on a testbed. This is expected to produce more significant results.