Ultra battery application for adaptive model predictive control in wind penetrated power systems

Abstract

This work pertains to utilization of small sized Ultrabattery (UB) in wind penetrated power system for LFC. Matlab level 2 S function coding is used to develop a custom made block for adaptive model predictive control for intelligent applications. Inner loop of UB is genetically tuned to mimic the first order reference system; the tuned storage system is connected with wind penetrated power system; with its voltage loop coupled to a function NACE (new area control error) for. A Staircase disturbance is introduced in wind penetrated power system. Investigation studies carried in MATLAB SIMULINK Environment reflect significant improvement in frequency response and tie power deviation of the system. The small rated UB combined with inertial response from wind farms marks for profitable operation. UB voltage and power response is also detailed and power constraints on power electronic converter are maintained within limits.

Keywords

Adaptive model predictive control ultrabattery load frequency control area control error new area control error genetic algorithim

Introduction

With ever increasing demand for energy, coupled with depleting fossil fuel reserves, it is difficult to sustain and maintain this growth, without involvement of energy from renewable sources. The green energy being environmental friendly and a permanent source therefore, it should not be surprising that in developed nations this technology has been paid maximum attention. Time has reached that developing countries or for that matter any country in the world cannot afford to overlook power production without utilization of renewable energy to a large extent. With improvements in power electronic controllers the wind penetration level up to 30% is achieved in power industry (Dahiya et al., 2015). In year 2019 the global wind industry witnessed new installations surpassing the 60 GW milestones for only the second time in history, but COVID-19 effects on the market outlook are still to be counted (Global Wind Energy, 2019).

Increasing involvement of wind energy in the conventional systems and its eventual participation in LFC is obvious. The major drawback in a power system is lack of energy reserves apart from the inertia present within the rotor of generators. To overcome this many energy storage devices available in market can be utilized for active power support. At present, energy storage applications incorporate valve-regulated lead–acid (VRLA), nickel metal hydride (Ni–MH) and rechargeable lithium batteries, flywheel energy storage System (FESS), superconducting magnetic energy storage (SMES) and supercapacitor.

In comparison to other energy storage devices, the VRLA battery offers low starting cost, a settled assembling base, broad appropriation systems and high reusing proficiency. However, In any case the running expense of the VRLA battery is costly by virtue of its short service span. For energy storage applications, the battery must be worked under high-rate partial state-of-charge (HRPSoC) specifically, inside a 30%–70% condition state of charge (SoC). This is since the battery can’t address the required current when the state of charge is at 30%. Ultrabattery is primarily a combination of supercapacitor and lead acid battery. Its major advantage is functioning at Partial state of charge without sulphation. It was proposed by Dr L T Lam in CSIRO energy technology in Australia and was further commercialized by Japan based company Furakawa (Cooper et al., 2009; Fairweather et al., 2013; Furukawa & CSIRO, 2008; Furukawa et al., 2010; Lam and Furukawa, 2009; Lam et al., 2007; Ribeiro et al., 2001).

In this work inertial response of DFIG in an assembly of UB with its power converter monitored by adaptive model predictive control (AMPC) strategy for intelligent application is studied. A 20% wind penetrated power system is found suitable for optimal performance. The penetration ratio is varied by changing the wind generators inertia constant. Greater value of inertia constant results in lesser wind penetration as given in Ribeiro et al. (2001). The power equilibrium point is determined by two functions as given in Huang et al. (2016), $Δ P_{f}^{*}$ the extra set point dependent on variations in frequency and $Δ P_{ω}^{*}$ on maximum turbine speed in relation to wind velocity and is governed by the following equations:

{Δ P}_{f}^{*} = - {k_{d}}_{f} \frac{df}{dt} - {k_{p}}_{f} Δ f

(1)

{Δ P}_{ω}^{*} = - {k_{w}}_{p} (ω^{*} - ω) + k_{wi} \int (ω^{*} - ω) dt

(2)

{Δ P}_{f ω}^{*} = {Δ P}_{f}^{*} + {Δ P}_{ω}^{*}

(3)

where k_df and k_pf represent the derivative gains for the derivative and proportional controllers respectively. Considering the above two cases of DFIG, ${Δ P}_{ω}^{*}$ varies comparatively slowly in comparison to the ${Δ P}_{f}^{*}$ frequency deviation power reference point. Therefore, in this situation during the time of perturbance in the system the angular frequency deviation could be approximated to zero. Further considering reference power point conversion instantly by converter the above equations could be equated to:

Δ P_{ren} = {Δ P}_{fw}^{*} = {Δ P}_{f}^{*} = - k_{df} \frac{df}{dt} - k_{pf} Δ f

(4)

Figure 1 represents the transfer function based design of conventional as well as the renewable system configured on DFIG based inertial emulation module as depicted in Huang et al. (2016).

Earlier works (Abo-Elyousr, 2016; Chang-Chien et al., 2011; Ibraheem et al., 2015; Jalali and Bhattacharya, 2013; Ma et al., 2017; Mohamed et al., 2012; Oshnoei et al., 2018; Sharma et al., 2014; Yingcheng and Nengling, 2012; Zhang et al., 2017) are considering use of interfacing inertial response of DFIG with various optimization techniques and controllers for damping electromechanical oscillations. In Dar and Mufti (2017) DFIG with one step ahead predictive configuration an SMES device is proposed. In this work a small sized Ultrabattery with comparatively less cost than SMES and a faster ramping response and pragmatic and practical approach is possible as Ultrabattery does not require any specific temperature conditions no dewar arrangement and cooling system is required. This paper is divided into five sections the UltraBattery model and its power conditioning system is discussed in section II. Simulation model represented in section II. The adaptive predictive scheme is illustrated in section IV. Case studies and effectiveness of this scheme is detailed in section V and conclusion is presented in section VI.

Ultrabattery disposition and its utilization in simulation studies

Ultrabattery technology modifies the chemistry of battery such that lead-acid battery provides power management and reduces negative plate sulfation. Basic arrangement of UltraBattery is shown in Figure 2.

Reduction of negative sulfation within the plates, which seems to be the chief source of ageing of VRLA batteries, is avoided in UB, as result of the carbon-configured supercapacitor connected in parallel and utilizing a similar electrolyte and common electrode as in VRLA. Ultrabattery technology is best adapted for delivering frequency control services to the grid which can be seen in Hajjam and Mufti (2021). It is ideal for PSOC activity which enables it to react in both directions by charging, unloading or adjusting the charging or unloading pace. It responds easily and can power up much faster than any traditional generator, tracks the control signal reliably and offers better support to the system operator. Public domain test data supporting this application was released in Ferreira et al. (2012), Hund et al. (2008a) and provide test periods representing regulatory services. Ecoult’s 3 MW deployments in Pennsylvania New Jersey Interconnection are an example of UB technologies conducting grid scale frequency control. The proposed UB equivalent circuit is shown in Figure 3. It is configured on AVR 95-33 developed at East Penn Ultrabattery (Ferreira et al., 2012; Hund et al., 2008b; Standard AVR-95-33, UltraBattery, 2017). The specifications of the device are given in Appendix ‘A’ and governed by the power converter interface linking the system with utility for bidirectional power exchange.

A state - variable model of UB can be obtained by defining state variables X_1, X₂ and X₃ as:

X₁ = Voltage across C_s

X₂ = Voltage across C_b

X₃= Voltage across C_x

From the above ckt it can be seen that,

I_{1} = I_{3} + I_{4}, I_{1} = I_{5} + I_{6}

(5)

I_{1} = \frac{X_{1}}{R_{t}} + C_{s} {\dot{X}}_{1}

(6)

I_{1} = \frac{X_{2}}{R_{d}} + C_{b} {\dot{X}}_{2}

(7)

I_{2} = C_{x} {\dot{X}}_{3}

(8)

V = I_{1} R_{i} + X_{1} + X_{2}

(9)

V = I_{2} R_{x} + X_{3}

(10)

Put equation (5) into (8), results in:

V = (\frac{X_{1}}{R_{t}} + C_{s} X_{1}) R_{1} + X_{1} + X_{2}

(11)

{\dot{X}}_{1} = - X_{1} (\frac{1 + \frac{R_{i}}{R_{t}}}{C_{s} R_{i}}) - \frac{X_{2}}{C_{S} R_{i}} + \frac{V}{C_{S} R_{i}}

(12)

Put equation (6) into (8), leads to

V = X_{2} \frac{R_{i}}{R_{d}} + C_{b} R_{i} {\dot{X}}_{2} + X_{1} + X_{2}

(13)

\dot{X_{2}} = - X_{1} [\frac{1}{R_{i} C_{b}}] - X_{2} [\frac{1 + \frac{R_{i}}{R_{d}}}{R_{i} C_{b}}] + \frac{V}{C_{b} R_{i}}

(14)

Put equation (7) into (10) we get

V = C_{x} R_{x} {\dot{X}}_{3} + X_{3}

(15)

{\dot{X}}_{3} = - \frac{X_{3}}{R_{X} C_{X}} + \frac{V}{R_{X} C_{X}}

(16)

In matrix form the given equation is written as:

[\begin{matrix} \dot{X_{1}} \\ \dot{X_{2}} \\ \dot{X_{3}} \end{matrix}] = [\begin{matrix} - [\frac{1 + \frac{R_{t}}{R_{X}}}{C_{S} R_{T}}] & - [\frac{1}{C_{S} R_{T}}] & 0 \\ - [\frac{1}{C_{b} R_{i}}] & - [\frac{1 + \frac{R_{i}}{R_{d}}}{C_{b} R_{i}}] & 0 \\ 0 & 0 & - [\frac{1}{R_{x} C_{x}}] \end{matrix}] \times [\begin{matrix} X_{1} \\ X_{2} \\ X_{3} \end{matrix}] + [\begin{matrix} \frac{1}{C_{S} R_{i}} \\ \frac{1}{R_{i} C_{b}} \\ \frac{1}{R_{X} C_{X}} \end{matrix}] \times V

(17)

The arrangement of UltraBattery interfaced in this study is represented by Figure 4. A buck boost converter is used for bidirectional power interface by varying the duty ratio from 0 to 1. A combination of transformer and inverter is maintained by the dc bus voltage whereas the converter at the line side regulates the dc bus voltage at 800 V. The converter power rating determines the rating of the device which is 100 kW and just 1% of total plant capacity. To tune the PI controller of the UB, genetic algorithm is used as represented in Figure 5. The power system used in the study is of third order ARX.

A process of converting block diagram into subsystems is utilized by varying the inertia constant as given in Ferreira et al. (2012). The DFIG with 20% wind penetration is considered. A staircase disturbance reflecting a load outage is introduced at an interval of 20, 80, 140 and 200 seconds. A function called new area control error (NACE) is utilized as reflected in Ferreira et al. (2012) this constraint imposed upon control plan assures the restoration of nominal voltage rating for the UB unit following a disturbance. In case of the nth control area, NACE given for the mth sampling sequence is represented by the following equation:

NACEn (m) = yn (m) = ACEn (m) + δ n (ACEn (m) - ACEn (m - 1) + λ (V_{UB} - {V^{*}}_{UB})

(18)

where, ACEn=Δfn +βΔPtien

The second term of equation (18) presents a variable which is proportional to the derivative of ACE. This term leads to a supervisory control which is efficient in damping out electromechanical oscillations. By δ and λ we introduce the weightages present in the derivative term and also deviation in UB voltage from nominal through NACE.

A controlled variable y(N) is a combination of the area control error ACE and actual voltage deviation function ΔV_UB as given in the equation below:

y (N) = NACE (nm) + k_{1} (1 - z^{- 1}) NACE (nm)

(19)

Three models are compared in the simulation studies first one is the conventional system with LFC, second with 20% wind penetration and third model is with 20% wind penetration plus UB with its power converter controlled by AMPC scheme for intelligent battery monitoring and dynamic restoration as can be seen from the simulation diagram represented in Figure 6.

Adaptive model predictive intelligent control

A word about this controller may be required in beginning

This model automatically adjusts online the parameters of its controller (Y), so as to maintain satisfactory performance when the parameters of plant u (controlled system) vary with time. Direct incorporation of physical constraints like power and energy on input and output states of the controlled system are taken into account. The receding horizon principle is employed wherein only the first element of the computed control sequence is applied to the plant and the other elements of control sequence are disregarded. A third order autoregressive exogenous (ARX) model will suffice the representation of the power system (Abbey and Joos, 2007; Gopal, 2008; Mufti et al., 2015; Żak, 2003; Zargar et al., 2017). The adaptive predictive scheme incorporates the online plant identification and future prediction of the values in one the fly.

A linear discrete time model for the controller design is given as:

Y (k) = \sum_{i = 1}^{N} α_{i} (y) (k - i) + \sum_{i = 1}^{N} β_{i} (u) (k - i) + ε (k)

(20)

In compact form,

Y (k) = θ (k) ϕ (k) + ε (k)

(21)

wherein Y(k) represents the controlled variable and u(k) control variable

The prior states of inputs and outputs are given by vector $θ (k)$ , this is used for prediction of parameters of $φ (k$ ), $ε (k)$ is white noise.

\overset{\land}{φ (k}) = [\begin{matrix} α_{1} & α_{2} & α_{3} & β_{1} & β_{2} & β_{3} \end{matrix}]^{T}

(22)

State variable formulation of adaptive predictive control is utilized and equation (20) is employed in below given observable canonical form:

X (K + 1) = AX (k) + Bu (k)

(23)

Y (k) = C X (k)

(24)

A = [\begin{matrix} α_{1} & 1 & 0 & \dots & 0 & 0 \\ α_{2} & 0 & 0 & \dots & 0 & 0 \\ ⋮ & ⋮ & ⋮ & ⋮ & ⋮ \\ ⋮ & ⋮ & ⋮ & ⋮ & ⋮ \\ ⋮ & ⋮ & ⋮ & ⋮ & ⋮ \\ ⋮ & ⋮ & ⋮ & ⋮ & ⋮ \\ ⋮ & ⋮ & ⋮ & ⋮ & ⋮ \\ ⋮ & ⋮ & ⋮ & ⋮ & ⋮ \\ α_{n - 1} & 0 & 0 & \dots & 0 & 1 \\ α_{n} & 0 & 0 & \dots & 0 & 0 \end{matrix}], B = [\begin{matrix} β_{1} \\ β_{2} \\ ⋮ \\ ⋮ \\ ⋮ \\ ⋮ \\ β_{n - 1} \\ β_{n} \end{matrix}] and C = {[\begin{matrix} 1 \\ 0 \\ 0 \\ . . . \\ 0 \end{matrix}]}^{T}

(25)

Proof

Let’s assume A B and C were known and X(k) and Y(k) are measurable The goal is to find the control sequence so that the desired control objectives are fulfilled. The objectives concerning the future behavior of the plant from next to current state up to a prediction horizon N_P. The concept of predictive control strategy is to minimize the cost function of the type:

J_{1} = \sum_{i = 1}^{N_{P}} [w (k + i) - \overset{\land}{y} (k + i)]^{2}

(26)

J_{2} = \sum_{i = 1}^{N_{P}} [w (k + i) - y (k + i)]^{2} + λ (u (k + i - 1)^{2}

(27)

where w(k+i) is the desired system output at the (k+1)^ith sampling instant.

$\overset{\land}{y}$ (k+1) is the predictive output at the (k+1)^th instant. The minimization of J with respect to reference power command represented by {u(K),u(k+1)………..u(k+N_P -1)} gives a sequence of future controls u(K), u(k+1)……..u(k+N_P-1). The first value of this sequence is applied to the plant. At the next sampling instant the receding horizon principle is repeated.

J_{1} = [w (k + 1) - \overset{\land}{y} (k + 1)]^{2} + [w (k + 2) - \overset{\land}{y} (k + 2)]^{2} . . . . . . . . . . [w (k + N_{P}) - \overset{\land}{y} (K + N_{P})]^{2}

(28)

In matrix form this can be written as :

J_{1} = {[\begin{matrix} w (k + 1) - \overset{\land}{y (k + 1)} \\ w (k + 2) - \overset{\land}{y (k + 2)} \\ ⋮ \\ ⋮ \\ w (k + N_{P}) - \overset{\land}{y (k + N_{P})} \end{matrix}]}^{T} \times [\begin{matrix} w (k + 1) - \overset{\land}{y (k + 1)} \\ w (k + 2) - \overset{\land}{y (k + 2)} \\ ⋮ \\ ⋮ \\ w (k + N_{P}) - \overset{\land}{y (k + N_{P})} \end{matrix}]

(29)

wherein ‘w’ does not depend on ‘u’ but ‘y’ depends on ‘u’ so we have to substitute value of y in terms of ‘u’

J_{1} = [W - \overset{\land}{y}]^{T} [w - \overset{\land}{y}]

(30)

where,

\begin{matrix} w = {[w (k + 1) w (k + 2) \dots \dots . w (k + N_{P})]}^{T} \\ \overset{\land}{y} = {[y (k + 1) y (k + 2) \dots \dots . y (k + N_{p})]}^{T} \end{matrix}

Expression for y (k+1) is given by following equations:

\overset{\land}{y} (k) = CX (k + 1)

(31)

\overset{\land}{y} (k + 1) = CX (K + 2)

(32)

\overset{\land}{y} (k + 2) = CAX (k + 1) CBU (k + 1)

(33)

To put the above equation in proper form we need to wrte X(k+1) in current state X(k)

Therefore

\overset{\land}{y} (k + 2) = CA [AX (k) + Bu (k)] + CBu (k + 1) = C A^{2} X (k) + CABu (k) + CBu (k + 1)

(34)

In matrix form it can be written as:

[\begin{matrix} \begin{matrix} \overset{\land}{y} (k + 1) \\ \overset{\land}{y} (k + 1 \\ ⋮ \\ ⋮ \\ ⋮ \end{matrix} \\ ⋮ \\ \overset{\land}{y} (k + N_{U}) \\ ⋮ \\ ⋮ \\ ⋮ \\ ⋮ \\ \overset{\land}{y} (k + N_{P}) \end{matrix}] J_{1} = [\begin{matrix} \begin{matrix} CA \\ C A^{2} \\ ⋮ \\ ⋮ \\ ⋮ \end{matrix} \\ ⋮ \\ C A^{N_{U}} \\ ⋮ \\ ⋮ \\ ⋮ \\ ⋮ \\ C A^{N_{P}} \end{matrix}] X (K) + [\begin{matrix} CB & 0 & . . . & 0 \\ CAB & CB & . . . & 0 \\ ⋮ & ⋮ \\ ⋮ & ⋮ \\ C A^{N_{u - 1}} B & C A^{N_{U - 2}} B & . . . & CB \\ ⋮ & ⋮ & ⋮ \\ ⋮ & ⋮ & ⋮ \\ C A^{N - 1} B & C A^{N - 2} B & . . . & \sum_{i = 0}^{N - N_{U}} C A^{i} B \end{matrix}] \times [\begin{matrix} u (k) \\ u (k + 1) \\ . \\ . \\ . \\ . \\ . \\ u (k + N_{U} - 1) \end{matrix}]

(35)

Here in short form it can be written:

Y = SX + LU

(36)

where Y is the short hand representation of $\overset{\land}{y} (k + 1)$ , $\overset{\land}{y} (k + 1) . . . . . \overset{\land}{y} (k + N_{P})$

\begin{matrix} y = [\begin{matrix} \begin{matrix} \overset{\land}{y} (k + 1) \\ \overset{\land}{y} (k + 1) \\ . \\ . \\ . \end{matrix} \\ . \\ \overset{\land}{y} (k + N_{U}) \\ . \\ . \\ . \\ . \\ \overset{\land}{y} (k + N_{P}) \end{matrix}] S = [\begin{matrix} \begin{matrix} CA \\ C A^{2} \\ ⋮ \\ ⋮ \\ ⋮ \end{matrix} \\ ⋮ \\ C A^{N_{U}} \\ ⋮ \\ ⋮ \\ ⋮ \\ ⋮ \\ C A^{N_{P}} \end{matrix}] L = [\begin{matrix} CB & 0 & . . . & 0 \\ CAB & CB & . . . & 0 \\ . \\ . \\ C A^{N_{u - 1}} B & C A^{N_{U} - 2} B & CB \\ . . . \\ C A^{N_{P} - 1} B & C A^{N_{P} - 2} B & \sum_{i = 0}^{N - N_{U}} C A^{i} B \end{matrix}]; \end{matrix}

(37)

\overset{\land}{y} = SX (K) + LU

(38)

Now the optimal control problem for a AMPC can be stated as:

minJ = \sum_{i = 1}^{N} {‖ y^{\land} (k + i) - w (k + i)] ‖}^{2} Q + \sum_{i = 1}^{N_{u}} {‖ u (k + i - 1) ‖}^{2} R 34

(39)

subject to the constraints being satisfied, where w is the desired value vector of system output, Q & R are weight matrices. Substituting (equation (5.13)) in (equation (5.14)) the optimization problem is reduced to a general quadratic form. In Matlab we can directly use ‘quadprog’ for solving this problem

\min J (U) = U^{T} HU + 2 G^{T} U

(40)

where,

\begin{matrix} s . t . A_{qp} U \leq bc \\ G = - [W - SX {(k)}^{T}] [Q] [S] and \\ H = {[L]}^{T} [Q] [L] + R \end{matrix}

(41)

Power limits of UB

In our UB controller design, factor Y is proposed to be a factor of ACE and voltage deviations, as different deviations, that is, tie-power, frequency and UB voltage, are needed to achieve value 0 in stable state. The inequality constraints A_qpU≤ bc imposed by small sized UB are implemented by following algorithm:

Sample the system output.

Estimate the model parameters.

Compute and apply the control signal.

Converter rating governs the maximum power limit for charging/discharging the UB. By keeping UB power command as control variable, converter rating forces the accompanying constraints on control vector u. Reference Power command which acts as control variable for UB shall not exceed the converter rating during energy releasing and storing. Thus to maintain limits on plant (UB), a control vector u needs to be incorporated as given below

FU \leq F

(42)

where F= $(\begin{matrix} I \\ - I \end{matrix})$ and U= $[\begin{matrix} {\vec{u}}_{\max} \\ {\vec{u}}_{\min} \end{matrix}]$ ${\vec{u}}_{\min}$ and ${\vec{u}}_{\max}$ are vectors.

Since the converter rating of UB is 1% of area capacity thus u_max = 0.01 p.u. and u_min = −0.01 p.u. The full form equation is:

[\begin{matrix} 1 & 0 & 0 & \dots & \dots & 0 \\ 0 & 1 & 0 & \dots & \dots & 0 \\ ⋮ & ⋮ \\ 0 & 0 & \dots & 1 & 0 & \dots \\ ⋮ & ⋮ \\ 0 & 0 & \dots & \dots & 0 & 1 \\ - 1 & 0 & 0 & \dots & \dots & 0 \\ 0 & - 1 & 0 & \dots & \dots & 0 \\ ⋮ & ⋮ \\ 0 & 0 & \dots & \dots & 0 & - 1 \end{matrix}] * [\begin{matrix} \begin{matrix} {P^{*}}_{UB} (k) \\ P^{*} UB (k + 1) \\ ⋮ \\ {P^{*}}_{UB} (k + N_{U}) \\ ⋮ \\ {P^{*}}_{UB} (k + N) \end{matrix} \end{matrix}] \leq [\begin{matrix} P_{max} \\ ⋮ \\ P_{max} \\ - P_{min} \\ ⋮ \\ - P_{min} \end{matrix}]

(43)

For implementing AMPC, a suitable prediction model of the system is required as detailed in Figure 7.

Simulation results and discussions

A continuous load disturbance is applied to the power system at 20,80 140 and 200 second interval as reflected in Plot of Figure 8. It is seen with respect to the step load disturbance the Ultrabattery stabilizes the system at the earliest and helps the power system to restore to steady state as seen from plot of Figure 9; wherein voltage level change, wind power change are detailed. In figure 10 plots of prediction estimates are depicted.

In comparison to the results reflected in Jalali and Bhattacharya (2013) and Dar and Mufti (2017) there is a significant improvement in the damping response of frequency and tie power deviations as reflected in the Table 1.

Table 1.

Comparison of various parameters.

Load disturbance	ΔP_D	Peak frequency deviation (Hz)		Peak tie power deviation (p.u.)
		Δf₁	Δf₂	ΔP_tie1
Without	0.01	0.03	0.03	0.014
With DFIG	0.01	0.016	0.016	0.007
With DFIG and SMES	0.01	0.014	0.014	0.003
% Reduction with DFIG	–	53.33	53.33	78.571
% Reduction with DFIG and UB	–	66.66	66.66	61.66

The simulation investigations carried out are discussed as follows:

The UltraBattery owing to its fast ramping rate and operation under HRPSoC has worked efficiently for suppressing electromechanical oscillations and achieving steady state for power system operation.

The small sized UB incorporated is just 1% of the total plant capacity hence it does not result in increased cost escalation of the overall wind penetrated power system.

The system is ready to face new disturbances and it is seen after tackling it, the UB comes back to its rated level of 480 V thus showing a very robust control performance.

Controller forces the converter to operate near its operating range of 0.01 p.u. At first disturbance it is touching this limit.

Power industry universally is going for green energy. Penetration of wind energy with conventional system is inevitable. Doubly fed Induction generators with inertial response capacity helps the power system in AGC operations. We propose an UltraBattery model, which is small sized hence cost effective, releases more kw/hr energy for its power size, compared to other energy storage devices in the market. UltraBattery helps in releasing or absorbing energy with respect to stochastic availability of power from wind turbines and at times of outages.

Conclusion

In this paper, AMPC for intelligent control of UB is recommended. We have used Matlab Simulink Environment by utilizing the level 2 S- function coding. Customized blocks are created and genetic tuning employed for UB power command tracking. A variable NACE new area control error is presented, which comprises combination of ACE and UB in feedback. This scheme forces the power converter to operate at its power limit of 0.01 p.u. The small sized UB does not escalate cost of wind penetrated power system. Incorporation of large data is needed for pragmatic and comprehensive studies wherein, factors like inequality constraints and external parameters of weather temperature and pressure are considered and all the dynamics associated with DFIG Wind Farms and load frequency control problem is addressed in a more pragmatic approach.

Footnotes

Appendix A

Appendix B

Appendix C

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

S Zahid Nabi Dar

References

Abbey

Joos

(2007) Teaching genetic algorithm using MATLAB. International Journal of Electrical Engineering Education 36(2): 139–153.

Abo-Elyousr

(2016) Load frequency controller design for two area interconnected power system with DFIG based wind turbine via ant colony algorithm. In: Eighteenth International Middle East Power Systems Conference (MEPCON), pp. 253-260, doi: 10.1109/MEPCON.2016.7836899.

Chang-Chien

Lin

Yin

(2011) Enhancing frequency response control by DFIGs in the high wind penetrated power systems. IEEE Transactions on Power System 26(2): 710–718.

Cooper

Furakawa

Lam

, et al. (2009) The UltraBattery—A new battery design for a new beginning in hybrid electric vehicle energy storage. Journal of Power Sources 188(2): 642–649.

Dahiya

Sharma

(2015) Optimal generation control of in terconnected power system including DFIG-based wind turbine. IETE Journal of Research 61: 285–299.

Dar

SZN

Mufti

(2017) Enhanced load frequency control response with integration of supervisory controlled superconducting magnetic energy storage system in wind-penetrated two-area power system. Wind Engineering 41(5): 330–342.

Fairweather

Stone

Foster

(2013) Evaluation of UltraBattery™ performance in comparison with a battery-supercapacitor parallel network. Journal of Power Sources 226: 191–201.

Ferreira

Baca

Hund

, et al. (2012) Life Cycle Testing and Evaluation of Energy Storage Devices. Albuquerque, NM: Sandia National Laboratories. Available at: www.sandia.gov/ess/docs/pr_conferences/2012/papers/Friday/Session1/03_Ferreira_PeerReview_Print.pdf

Furukawa & CSIRO (2008) Development of UltraBattery®: 3rd report. Furukawa Electric Company Ltd. Available at: www.furukawadenchi.co.jp/english/research/new/pdf/ultra_03.pdf

10.

Furukawa

Takada

Monma

, et al. (2010) Further demonstration of the VRLAtype UltraBattery under medium-HEV duty and development of the flooded-type UltraBattery for micro-HEV applications. Journal of Power Sources 195(4): 1241–1245

11.

Global Wind Energy Report 2019 Global Wind Energy Council Available at : https://gwec.net/global-wind-report-2019

12.

Gopal

(2008) Digital Control and State Variable Methods. Noida, India: Tata McGraw Hill.

13.

Hajjam

Mufti

M-D

(2021) An intelligent two-level control for improved automatic generation control of a multisource deregulated power system. International Journal of Energy Research 45(5): 1–28.

14.

Huang

P-H

Kuo

J-K

Huang

C-Y

(2016) A new application of the Ultra Battery to hybrid fuel cell vehicles. International Journal of Energy Research 40(2): 146–159.

15.

Hund

Clark

Baca

(2008a) UltraBattery® test results for utility cycling applications. In: The 18th international seminar on double layer capacitors and hybrid energy storage devices. Fort Lauderdale, FL: Redox Engineering, LLC. Available at: www.ecoult.com/wp-content/uploads/2012/03/Hund-UltraBattery®-Utility-Testing-12-08.pdf

16.

Hund

Clark

Baca

(2008b) Ultrabattery test results for utility cycling applications. SAND2008-7451C. Power Source Component Development Department, Sandia National Laboratories, Albuquerque, NM.

17.

Ibraheem Niazi

Sharma

(2015) Study on dynamic participation of wind turbines in automatic generation control of power systems. Electric Power Components Systems 43(1): 44–55.

18.

Jalali

Bhattacharya

(2013) Frequency regulation and AGC in isolated systems with DFIG-based wind turbines. In: IEEE power and energy society general meeting (PES), Vancouver, BC, Canada, 21–25 July.

19.

Lam

Furukawa

(2009) Secondary Batteries – lead – acid Systems | Supercap Hybrid (UltraBattery™). In: Encyclopedia of Electrochemical Power Sources, pp.755. Elsevier Science & Technology.

20.

Lam

Louey

Haigh

, et al. (2007)VRLA ultrabattery for high-rate partial-state-of-charge operation. Journal of Power Sources 174(1): 16–29.

21.

Liu

Zhang

(2017) LFC for multi-area interconnected power system concerning wind turbines based on DMPC. IET Generation, Transmission & Distribution 26(2): 2689–2696.

22.

Mohamed

Morel

Bevrani

, et al. (2012) Model predictive based load frequency control design concerning wind turbines. International Journal of Electrical Power & Energy Systems 43(1): 859–867.

23.

Mufti

Iqbal

Lone

, et al. (2015) Supervisory adaptive predictive control scheme for supercapacitor energy storage System. IEEE Systems Journal 9(3): 1020–1030.

24.

Oshnoei

Khezri

Muyeen

, et al. (2018) On the contribution of wind farms in automatic generation control: review and new control approach. Applied Sciences 8(10): 1848.

25.

Ribeiro

Johnson

Crow

, et al. (2001) Energy storage systems for advanced power applications. Proceedings of the IEEE 89(12): 1744–1756.

26.

Sharma

Niazi

Ibraheem , et al. (2014) LS-SVM based AGC of power system with dynamic participation from DFIG based wind turbines. In: Proceedings of the 3rd renewable power generation conference (RPG), Naples, Italy, 24–25 September 2014, pp.1–6. doi: 10.1049/cp.2014.0927.

27.

Standard AVR-95-33, UltraBattery (2017) Technical Specifications of UBer™ USB-480 V UltraBattery Storage Block. https://www.parliament.nsw.gov.au/ladocs/other/8056/CSIRO%20attachment%201a%20-%20H

28.

Yingcheng

Nengling

(2012) System frequency regulation in doubly fed induction generators. International Journal of Electrical Power & Energy Systems 43(1): 977–983.

29.

Żak

(2003) System and Control. New York, NY: Oxford University Press.

30.

Zargar

Mufti

MU-D

Lone

(2017) Adaptive predictive control of a small capacity SMES unit for improved frequency control of a wind diesel power system. IET Renewable Power Generation 11(14): 1832–1840.

31.

Zhang

Liu

(2017) Distributed model predictive load frequency control of multi-area power system with DFIGs. IEEE/CAA Journal of Automatica Sinica 4(1): 125–135.