- Email: [email protected]
Value of gas network infrastructure flexibility in supporting cost effective operation of power systems
Applied Energy 202 (2017) 571–580
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Value of gas network infrastructure flexibility in supporting cost effective operation of power systems Hossein Ameli a,⇑, Meysam Qadrdan b, Goran Strbac a a b
Control and Power Group, Imperial College, London, United Kingdom Institute of Energy, Cardiff University, Cardiff, United Kingdom
h i g h l i g h t s Integration of renewable energies causes operational challenges in power systems. Value of operational practices for gas and electricity networks were quantified. Role of gas network flexibility to address the challenges were evaluated. Less gas load curtailment when the gas network is more flexible. Cost savings when integrated approach and gas network flexibility were considered.
a r t i c l e
i n f o
Article history: Received 28 March 2017 Received in revised form 3 May 2017 Accepted 18 May 2017
Keywords: Integrated gas and electricity networks Wind generation variability Economic dispatch Flexibility Multi-directional compressors
a b s t r a c t The electricity system balancing is becoming increasingly challenging due to the integration of Renewable Energy Sources (RES). At the same time, the dependency of electricity network on gas supply system is expected to increase, as a result of employing flexible gas generators to support the electricity system balancing. Therefore the capability of the gas supply system to deliver gas to generators under a range of supply and demand scenarios is of a great importance. As potential solutions to improve security of gas and electricity supply, this paper investigates benefits of employing flexible multi-directional compressor stations as well as adopting a fully integrated approach to operate gas and electricity networks. A set of case studies for a GB gas and electricity networks in 2030 have been defined to quantify the value of an integrated operation paradigm versus sequential operation of gas and electricity networks. The results indicate there are significant overall system benefits (up to 65% in extreme cases) to be gained from integrated optimization of gas and electricity systems, emphasizing the important role of gas network infrastructure flexibility in efficiently accommodating the expected expansion of intermittent RES in future power systems. Ó 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http:// creativecommons.org/licenses/by/4.0/).
1. Introduction The share of Renewable Energy Resources (RES) is projected to increase significantly across the Great Britain (GB) network to meet the de-carbonization target in the future [1]. The installed capacity of wind power generation is expected to be between 52 GW [1] and 64 GW [2] in 2030 GB power system. This increase in wind generation share leads to balancing challenges due to the variable nature of the wind. In this context, flexible gas-fired generation with fast ramping rates is expected to facilitate cost effective integration of wind generation to meet the balancing challenges of ⇑ Corresponding author. E-mail addresses: [email protected] (H. Ameli), [email protected] (M. Qadrdan), [email protected] (G. Strbac).
supply and demand. The interdependency of the gas and electricity networks is increasing because of the role of gas-fired plants as the main linkage of these networks. Thus, the wind variability is reflected as intermittency in the gas demand as well. Based on higher inertia of gas, delivery time from supply sources (terminals and storage facilities) to demands in the gas network is much higher than electricity transmission. Hence, the Gas System Operator (GSO) maintains minimum volumes of gas, called Linepack, within the pipelines to meet rapid changes in the network. The variability of gas demand for power generation, consequently influences the linepack and makes its management more difficult. Real operational data from National Grid indicates that the linepack of the GB high pressure gas network in 2012 fluctuated with larger magnitude compared to year 2002 [3].
http://dx.doi.org/10.1016/j.apenergy.2017.05.132 0306-2619/Ó 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
572
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Nomenclature
Parameters & variables C cost (£) CR maximum compressor pressure ratio e produced green house gas emissions (tonnes) F objective function (£) N set P electrical power (MW) p gas pressure (Pa) p upper pressure bound (Pa) p lower pressure bound (Pa) Q volumetric gas flow rate (m3 =s) T time horizon (h) s time step (= 1 h) l compressor fuel rate coefficient (0.084 m3 =MJ) c polytropic efficiency (4.70 MJ=m3 ) g efficiency (%) f amount of gas trapped by a compressor (m3 =s) Superscripts a!b from node a to node b comp compressor dis discharge ecomp electrically-driven compressors elec electricity eload electrical demand em emission eshed electrical load shedding
Several studies have examined the role of flexibility options in addressing the balancing challenges driven by RES in power systems [4–8]. Flexibility is described as the ability of the system to deal with uncertainty and variability in the network, while keeping an acceptable level of reliability, from time scales of seconds to hours [4]. In [5] a framework considering generation flexibility requirements and environmental constraints for the Unit Commitment (UC) process is proposed. It was concluded, through utilizing flexible plants the occurrences of supply and demand imbalances caused by the wind forecast errors was decreased. In [7] it is shown that through using multi-mode operation of Combined Cycle Gas Turbines (CCGTs) the system security was increased. The multimode operation of CCGTs is related to the units which incorporate a steam bypass stack. These units are capable also to operate in open-cycle mode, which increase the flexibility of these gas plants in order to deal with large integration of wind into the system. Although less efficiency was achieved when operating in opencycle mode, however it was shown that in overall the emissions were significantly reduced as the generation portfolio has been changed and the production of generation units characterized by high emissions is decreased. The focus of these studies was mainly on the impact of the flexibility options in power systems operation, while the integrated energy systems value (gas and electricity) of flexibility has not been taken into account. Although, the interaction of gas and electricity networks have been studied from security of supply perspectives [9], wind forecasting and uncertainty [10,11], energy adequacy [12], coordinated scheduling [13,14], market coupling [15], and also planning strategy [16], however comprehensive analysis of large penetration of RES on the gas network operation has been reported by couple of studies [12,17–20]. The value of an integrated operation practice of the gas and electricity networks in comparison to sequential operation of these networks, has not
gload gas demand gshed gas load shedding int integrated pgen gas for power generation pumpinj injection to the pump storage unit pumpwith withdrawal from the pump storage unit sd shut-down storinj injection to the storage facility storwith withdrawal from the storage facility su start-up suc suction supp supply wavail available wind wcur curtailed wind var variable Subscripts b electrical busbar c compressor station g power generation unit k thermal power generation unit m gas node l gas pipeline p pump unit s gas storage facility t time y gas terminal
been evaluated in these studies. Furthermore, in the literature, electricity side flexibility options are discussed and to the best of our knowledge the value of gas network flexibility has not been investigated. In a few studies, including [21–23] the planning and operation optimization of the gas transmission networks from different aspects are studied. Additional pipelines and compressor stations are the main reinforcement components in the planning mode of these studies. In [24], an assessment on the role of gas network infrastructure in providing energy security has been studied. But, none of these studies considered the role of multi-directional compressor units in operation of the gas network. Although the role of multi-directional compressor as a mean of providing flexibility to the gas network is discussed in [25], however the research was focused on the gas network infrastructures and the value of this flexibility option was not quantified. As mentioned above, the future growth of wind generation capacity will increase the complexity of gas network management. One of the solutions to deal with these challenges is to make the gas network infrastructures flexible. Gas storage facilities and flexible multi-directional compressor stations are potential candidates to cope with the growing variabilities in the gas network. The interest of the GB system operator to invest in more flexible stations able to compress gas in different directions is reported in [26]. For example, if the available gas in North Sea could not be delivered to the network, one-directional compressor stations are not able to compensate for the lack of gas supply to meet the demand. Thus, redirecting the gas flow through the compressors could help to avoid additional reinforcement in the network. In this paper, potential solutions for improving capability of gas and electricity supply system to meet the demands were investigated. The solutions are (a) changing the operation practice through adopting an integrated approach to operate gas and electricity networks, (b) employing flexible multi-directional
573
H. Ameli et al. / Applied Energy 202 (2017) 571–580
compressor stations in the gas transmission network, and (c) the first two solutions combined. In the operation of gas and electricity networks, the multi-directional compressors are modeled so that the gas flow could be redirected to ensure that the gas demand is met while minimizing the operational cost of the gas and electricity networks. The value of multi-directional compressors were especially highlighted, when large variations in the gas network was happened (e.g., post-contingency conditions). Therefore, a set of gas supply interruption scenarios for a GB gas and electricity network in 2030 were defined. The efficacy of the above solutions to mitigate the impacts of the gas supply interruptions on the operation of gas and electricity networks were quantified. In these scenarios as a contingency was happened in the gas supply system, difficulties in the operation of the gas network was occurred. Due to the fact that location of majority of the gas plants were not very close to the gas supply points, gas might not be delivered in time due to limitations in speed in transporting gas through gas transmission grid. In order to compensate for limited electricity production by gas plant, more expensive options were required to produce more and imports from interconnection were increased. On the other hand, as the flexibility in the gas network was enhanced through multi-directional gas compressors, gas delivery to the gas plants was possible. Consequently, the overall emission and the total operational costs of the networks were reduced. 2. Modeling methodology 2.1. Integrated gas and electricity networks The enhanced version of the Combined Gas and Electricity Networks (CGEN) operation model [18] was used in order to represent a detailed model of flexible multi-directional compressor stations. This model was employed to examine the role of the gas network infrastructure flexibility in supporting future power systems with significant penetration of wind generation. The model is able to optimize the operation of the gas and electricity networks, simultaneously. Constraints of the power system modeled including characteristics of thermal generating units (i.e. ramp up/down, minimum up/down time, minimum stable generation of thermal units) and spinning reserve requirements, and energy balance (1) were considered. In the supply side of the energy balance, generation of the units, net energy of pump storage, absorbed wind by the grid are taken into account. On the other side (i.e., demand), electricity demand and electricity required for electrically-driven compressors are considered. The electrical load shedding term prevents the model from infeasibility, which shows the Value of Lost Electricity Load (VoLEL) in the electricity network. Eq. (1), indicates that at each time step (i.e., hourly) the supply and demand has to be balanced.
8t 2 T : ¼
8x 2 Nm ; t 2 T :
Q supp s þ Q x;t s þ ðQ storwith Q storinj Þs x;t x;t x;t
þ ðQ comp fx;t Þ s ¼ Q gload s þ Q pgen s Q gshed s x;t x;t x;t x;t
ð2Þ
Gas flow dynamics associated with linepack changes were captured in this granularity level. A quasi steady state condition for gas flow over a day was assumed. The intra-day variations of gas supply and demand were taken into possession in hourly time steps [27]. Detailed description and formulations of gas and electricity network operational modeling is presented in [20]. Gas-fired generating plants as well as electrically-driven compressors are the coupling components of electricity and gas networks. In electricity network, gas plants are electricity providers, while in gas network, gas-fired generating plants consume gas (i.e., gas load). On the other hand, electrically driven compressors consume electrical power (i.e., electrical load) and maintain the gas network pressure. An hourly time step rolling planning approach was applied for the network modeling. The main reasons for implementing a rolling planning approach was to enhance efficiency of computations by splitting into seven smaller-size optimization problems (from a week to a day duration). In the modeling, the electricity network states (e.g., ON/OFF state of the thermal generating units) of last hours of the previous day was taken into account for the next day. In the gas sector, based on the linepack and storage of the previous day the state of the gas network for the next day was updated. 2.1.1. Flexible multi-directional compressor stations To recover the loss of pressure in the gas infrastructure, which is caused by friction in the pipelines, installation of compressor units are considered. Compressors operate either electrical or gas driven. The compressor prime mover required power is presented in [27]. In order to model the multi-direction option of the compressor stations, the suction and discharge nodes of the compressor should be decision variables and not pre-set. Therefore, from the modeling point of view, it was assumed that at each compressor station node, virtually two compressor units are installed (from node a ! b, and b ! a (Fig. 1)), which compress the gas in opposite directions. At each time step only one of these compressors could operate. Through calculating the power consumption of these reverse compressors (3) and (4), the one with more power consumption (5) will dictate the direction of the gas flow at each time step. This means that the pressure in discharge node of the optimal compressor is more than the suction node. The pressure ratio of the discharge (outlet) node to suction (inlet) node is limited to CR (6). 8c 2 Nc ; t 2 T:
Ng Np Nb X X X Pg;t s þ ðP pumpwith Ppumpinj Þsþ Pwind s p;t p;t b;t g¼1
Nb X
ity energy balance, a gas load shedding term is considered in the gas flow balance which presents the Value of Lost Gas Load (VoLGL).
p¼1
ðPeload s Peshed s þ Pecomp sÞ b;t b;t b;t
b¼1
ð1Þ
b¼1
In the gas network modeling, constraints such as gas supply limits for terminals and storages, linepack changes, pressure limits, and gas compressor operation limits, and gas flow balance (2) were taken into account. The gas flow balance should be satisfied at each node and each time step. Gas provided through terminals, gas available in the pipelines, net gas from the storage facilities, and gas from the compressors excluding the gas consumed by gasdriven compressors are the key terms in the supply side. Two type of demands are considered; (a) required gas for the plants for power generation, and (b) gas for other uses. Same as the electric-
Fig. 1. A simplified structure of a flexible multi-directional compressor station.
574
H. Ameli et al. / Applied Energy 202 (2017) 571–580
2
Pa!b ¼ c;t
c:Q comp pbc;t c;t :4 a comp g pc;t 2
Pb!a ¼ c;t
c:Q comp pac;t c;t :4 b comp g pc;t
3
!ð1=cÞ
15
ð3Þ
3
!ð1=cÞ
15
ð4Þ
a!b Pcomp ¼ maxðPc;t ; Pb!a c;t c;t Þ
ðCRÞ
1
pac;t
6
pbc;t
6 CR
ð6Þ
Q comp 6 Q comp c;t c
ð7Þ
Pcomp c;t
ð8Þ
6
Pcomp c
pcomp c
ð9Þ
P pcomp c
ð10Þ
6
fc;t ¼ l:Pcomp c;t
ð11Þ
Flexible multi-directional compressor stations are able to flow the gas in different directions. For this purpose, it is required to reinforce the existing compressor units with additional piping and valves in the station yard. Currently, the GSO in GB is using Multi-junctions (i.e., pipelines, valves, and regulators) to control the flow of gas in the National Transmission System (NTS). Multijunctions can be located close to compressor stations, and therefore this make the gas compression in different directions possible [28]. Once the compressor becomes multi-directional, the gas demand will dictate the direction of the gas flow in the network [25]. This will enhance the gas network infrastructure flexibility, which improves the action of GSO for meeting the consumers requirements. A simplified structure of a flexible multidirectional compressor station including additional valves and pipelines are presented in Fig. 1. The red1(small dashed) line and the blue (dashed) line indicate the ability of modifying the gas flow in reverse directions. At each time the valves of the opposite direction are closed. The proposed model is able to change the direction of the gas flow in the compressors through comparing the pressures of different junctions in the station yards. 2.1.2. Objective functions In the electricity network, costs of power generation, electrical load shedding, and CO2 emission penalties were taken into account in the objective function (12). In the gas network the costs of gas supplies, storages, changes of the gas flow, and gas load shedding were considered (13). The objective function of the Integrated modeling was to minimize the costs of both networks (14).
F elec ¼
Ng Nk T X X X var sd em ðC fuel þ C Þ:P s þ ðC su g;t g g k þ C k þ C k :ek;t sÞ t¼1
g¼1
Nb X C eshed :Peshed s þ b;t
k¼1
!
ð12Þ
b¼1
F gas ¼
Ny Nl T X X X C gas :Q supp sþ C gas :@Q l;t s y;t t¼1
þ
y¼1
In Integrated approach, in order to avoid twice calculation of gas plants generation cost, C fuel of gas power plants are zero, as the gas consumption of these plants were considered as a type of demand in the gas network operation.
2.2. Integrated vs. Sequential strategy 2.2.1. Sequential modeling In the Sequential mode, the operation of electricity network (including UC/Economic Dispatch and power flow) was optimized and the gas demand for power generation was then calculated. In this approach, the gas demand for power generation was provided as an input to the gas network operation model. In this modeling, the operational costs of the electricity network were minimized (12) and afterwards the operational costs of the gas network were minimized (13). The algorithm of the Sequential modeling using a rolling planning approach is presented in Fig. 2. The red (dashed) line indicates the sequential link of the electricity and gas networks. The Sequential approach does not consider the gas supply constraints while optimizing operation of the power system. Hence, a key disadvantage of this modeling is that potential interruptions in the gas supply do not influence the dispatch of gas-fired plants in the electricity network, as the modeling of electricity system is carried out before the gas network.
2.2.2. Integrated modeling Unlike the Sequential mode, in the Integrated modeling the gas and electricity network constraints were considered synchronously. Fig. 3 illustrates the algorithm of the Integrated modeling. The Integrated model minimizes the total operational cost of the gas and electricity networks, simultaneously (14). The corresponding optimization problem of Integrated model was solved through Mixed Integer Non-Linear Programming (MINLP) approach in Xpress-IVE (i.e., mmnlp module) [29]. The total number of variables for the week is about 260 k variables including about 20 k integer variables. However, as mentioned previously, a rolling approach has been implemented and the optimization is divided into seven smaller optimization problems. The main reason of using a rolling approach is that due to the non-linearity as well as integer space of the model, by increasing the number of variables, the computational time would be increased exponentially. In order to get more accuracy for modeling each day, four hours of the next day is also considered. As a result, at each step of the rolling approach around 43 k variables including around 3500 integer variables has been optimized. A branch and bound strategy has been used to find the solution of the binary variables. The search has been stopped when the absolute tolerance of the best solution’s objective function and the current best solution bound is less than a set value which is presented in (15).
jF BestBoundj 6 0:1% maxðF; BestBoundÞ
l¼1
Ns Nm X X ðC storwith :Q storwith C storinj :Q storinj Þsþ C gshed :Q gshed s s;t s;t m;t s¼1
ð14Þ
ð5Þ
Each compressor was subjected to maximum flow rate, power consumption, and capacity constraints (7)–(10). The tapped gas was calculated through (11). 8c 2 Nc ; t 2 T:
pdis c;t psuc c;t
F int ¼ F elec þ F gas
!
m¼1
ð13Þ 1 For interpretation of color in Figs. 1 and 2, the reader is referred to the web version of this article.
ð15Þ
The model was performed on a 3.20 GHz Intel(R) Xeon(R) processor and 16 GB of RAM system. The simulation time for different case studies was between 93 and 118 min. Initial values based on output of the optimization for a single time step were given to all decision variables. In addition, the decision variables were scaled so that the order magnitude of the values do not vary significantly. These two procedures aided to obtain convergence and stability of the optimization process.
575
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Fig. 2. Sequential mode algorithm.
Base (Case 1): In this case the Sequential modeling of gas and electricity networks (see Fig. 2) was considered. Direction of gas flow through compressors (i.e. suction and discharge nodes of compressors) were pre-set. Flexible multi-directional compressor stations (Case 2): The compressors were assumed to have the capability to change the direction of gas flow (see Fig. 1). The suction and discharge nodes of each compressor were decision variables that were determined as part of the optimization problem of the gas network operation. Integrated operational practice (Case 3): In this case the operation of gas and electricity networks were optimized simultaneously (see Fig. 3). The directions of gas flow through compressors were pre-set. Fully flexible (Case 4): Both solutions presented in Case 2 and Case 3 were combined in this case study.
Fig. 3. Integrated mode algorithm.
Impacts of partial and complete outage of Bacton gas terminal, the largest gas terminal in GB, on the operation of the gas and electricity networks were investigated for different cases. Two outage scenarios were defined: (a) interruption of gas import from Balgzand to Bacton Line (BBL) and Interconnectors UK (IUK) and (b) interruption of gas supply from whole Bacton terminal (see, Fig. 4). These studies were to assess how Integrated approach along with flexible multi-directional compressor stations could mitigate the adverse effect of an incident in the gas network.
3. Case studies 3.1. Descriptions of the cases As it was mentioned previously, three solutions were suggested in this paper. Changing the operation practice (i.e., integrated approach of gas and electricity modeling), reinforcing the physical assets of the gas network (i.e., flexible multi-directional compressor stations), and combination of these two. To evaluate the capability of the solutions in improving the security of supply, following cases for the GB electricity and gas networks were examined:
Fig. 4. Bacton supply sources.
576
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Fig. 5. GB 29-Busbar electricity transmission network. Fig. 6. GB simplified NTS gas network.
3.2. GB gas and electricity 2030 networks A GB electricity transmission network including 29 busbars, 47 transmission lines and 148 generators based on [26] was implemented in this research (Fig. 5). An updated version of the simplified NTS network (63 nodes, 54 gas pipelines, 9 terminals, and 5 storages), introduced in [17] based on [3] was applied in the model (Fig. 6). 3.3. Input data
Table 1 Generation mix in GB 2030 [1]. Technology
Capacity (GW)
Wind Gas Interconnector Nuclear Coal with CCS Pumped storage Hydro Other
52 33 11.5 9 4.5 2.7 1.1 1.2
The GB gas and electricity networks were modeled over a typical winter week in 2030. Table 1 presents the power generation mix in this study (see Fig. 7). In the National Grid outlook in 2030, 52 GW wind generation is projected to be installed across GB [1]. The hourly wind generation, electricity demand data as well as non-electric gas demand data in 2030 were taken from [20]. It is worth noting that in the model wind generation data was defined as available wind power. The available wind power could be absorbed by the electricity system or had to be curtailed due to the excess of wind generation (16).
8b 2 Nb ; t 2 T :
Pwavail ¼ P wind þ Pwcur b;t b;t b;t
ð16Þ
Fig. 7. Hourly available wind power, gas demand, and electricity demand [20].
577
H. Ameli et al. / Applied Energy 202 (2017) 571–580
4. Results and discussion 4.1. Partial outage of Bacton gas terminal 4.1.1. Generation mix The operation of the electricity network in the Sequential cases (Case 1 and Case 2) were the same, as it was optimized irrespective to the operation of the gas network. In Integrated Cases (Case 3 and Case 4) the constraints of gas supply were taken into account. Changes in electricity generation by different technologies in Integrated cases compared to Sequential cases are presented in Fig. 8. The generation of nuclear and hydro did not vary considerably as these are lowest cost generation technologies [17]. Moreover, hydro capacity was relatively small (see Table 1), so these plants did not make significant contribution to meeting electricity demand. Total generation of gas plants over the week in Integrated cases was less than Sequential cases. The reason is that in the Integrated modeling the gas and electricity networks operation were optimized simultaneously, therefore gas supply constraints were considered in the economic dispatch of power generation. As a result, roughly 18,000 and 7,000 tonnes of CO2 production were reduced in Case 3 and Case 4 (compared to Sequential cases), respectively. Due to the flexible multi-directional compressor stations option, gas flow was redirected in different time steps in order to deliver gas from different supply points. Therefore, because of the provided flexibility in the gas network the gas plants in Case 4 generated about 40 (GWh) more than Case 3 over the time horizon. In Case 3 in order to meet the energy balance, about 35 (GWh) electrical energy was imported through interconnectors. Whereas, in Case 4, the power flow through interconnectors has been decreased. In Fig. 9-a, the role of gas plants as an economical option [17] able to complement the lack of wind generation during low-wind, high-demand period (105–118 h) is presented. In Sequential cases, the gas supply constraints were not considered in operation of the electricity network. Therefore, as the electricity network operation was modeled first (see Fig. 2), it was preferred operating gas plants rather to importing energy via interconnection to keep the energy balance in the network, as this minimizes the operation costs. On the other side in Integrated cases as gas supply interruption at Bacton terminal was taken into account in the operation of power system, the generation through gas plants was reduced compared to the Sequential cases. This was mostly due to the fact that less gas was available close to Bacton terminal, which has affected especially the gas plants in busbars 20 and 26. In Case 4, thanks to the gas flow redirection option, delivering gas to the aforementioned plants could be managed cost effectively. The reduced generation from gas plants in Case 3 and Case 4 were compensated with interconnectors. Fig. 9-b presents the contribution of interconnectors in providing the energy balance in low-wind, high-demand period. It is worth noting that although the price of importing energy via interconnectors was assumed to be higher
Fig. 8. Electricity generation changes in respect to Sequential cases.
Fig. 9. Power supplied during low-wind, high-demand periods: (a) Gas-plants, (b) interconnectors.
than gas plants [17], the interconnection prevented gas load shedding in the gas network. Whereas, in the Sequential cases the model selected seemingly lower cost option based on using gas plants rather interconnector, as the real-time gas availability was not considered.
4.1.2. Gas supply and linepack Total gas supply through terminals and gas storage facilities are presented in Table 2. Changing the operational practice as well as the provided flexibility improved the gas delivery to the demand centers by reducing the gas transmission congestion. Fig. 10 shows the linepack in the pipelines during the simulation horizon. The linepack on a particular day of the week changes based on the demand pattern for power generation on that day. The linepack fluctuations was related to the fact that in off-peak times (e.g. in the mornings before 06:00) the gas demand was low, and therefore linepack was increasing. In the peak-times
Table 2 Total gas supply over the week. Case study
Case 1
Case 2
Case 3
Case 4
Gas supply (mcm)
2285.8
2322.5
2311.6
2319.1
Fig. 10. Linepack in different cases.
578
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Table 3 Total gas load shedding (Bacton partial outage). Case study
Case 1
Case 2
Case 3
Case 4
Gas for power generation (mcm) Gas for other uses (mcm)
5.62 29.69
0.85 2.27
0.00 0.00
0.00 0.00
(e.g. 17:00–20:00) the stored gas in the pipelines was used and the linepack has been reduced. The analysis of the amount of the linepack indicates more availability in the within-pipeline storage capability of the gas network in multi-directional cases (Case 2 and Case 4). This enables the gas network operators to manage rapid changes in the gas demand in short term. The gas load shedding in different cases is shown in Table 3. In Case 1 the total amount of gas load shedding throughout the week was about 35 million cubic meter (mcm) due to the low flexibility of the network. As explained before, in the Sequential cases, first the electricity network was modeled, and different states of the gas network were not considered. This means that the electricity network requested gas for generation without any constraints on gas supply. On the other side, as an outage occurred, the gas could not be delivered to the demands. In Case 2 there was flexibility in the gas network physical assets but the requested gas for power generation was still more than supply. Hence, in the modeling’s point of view about 0.85 (mcm) and 2.27 (mcm) of the gas for power generation and non-electric gas demand had to be curtailed, respectively. In Integrated modeling the above mentioned problem was obviated as the gas supply constraints were considered and so it was not required to shed any gas demand. A comparison between Case 1 and Case 3 indicates the importance of considering gas and electricity constraints synchronized as through proper scheduling of the gas plants there was no need to curtail gas demands. In the Sequential modeling, gas for power generation and other uses have the same value, and therefore there was no priority in curtailing different type of demands.
balancing issues. A positive power consumption means that the compressor was operating in the pre-set direction, whereas a negative value of power consumption indicates that the gas flow was redirected. In normal condition, the direction of the gas through compressors close to the gas terminals was same as the injected gas through terminals. Thus, when an incident in the gas terminals happened, compressor units close to the terminal stations had to modify the direction of the gas flow in order to meet the gas demand in areas close to the terminals and keep the gas pressure in the acceptable range as much as it was feasible. As an example, compressor station number nine (located in node 62, 63) mostly operated in the reverse direction. In addition, the power consumption by the compressors in Case 4 was less than Case 2 which means the gas network operation was managed better as it was less need to compress the pressure. 4.1.4. Gas and electricity networks costs Operational cost of the networks for various case studies over the typical winter week in partial outage of Bacton gas terminal are presented in Table 4. It is noteworthy that the penalty price of gas load shedding was assumed to be 11.1 £M=mcm [27]. The cost savings in different case studies are presented in Fig. 13. It is demonstrated that through operating simultaneously the gas and electricity networks and employing flexible multidirectional compressor stations, the lack of gas supply was completely compensated and about 30% cost saving in comparison to Case 1 was achieved.
Table 4 Operational costs in £M (Bacton partial outage). Case study Case Case Case Case
1 2 3 4
Electrical network
Gas network
Total
78.92 78.92 92.90 80.21
1190.66 834.49 808.18 810.39
1269.58 913.41 901.08 890.60
4.1.3. Compressor direction modification The proposed model was capable of changing the direction of the compressors. Mostly the compressors close to the Bacton gas terminal operated in reverse direction than the pre-set direction. In Figs. 11 and 12 it is shown that how flexible multi-directional compressor stations could provide flexibility to manage better the
Fig. 13. Cost savings in comparison to Case 1.
Fig. 11. Compressors power consumption in Case 2.
Fig. 12. Compressors power consumption in Case 4.
Fig. 14. Gas load shedding (Locations).
H. Ameli et al. / Applied Energy 202 (2017) 571–580 Table 5 Gas load shedding (Outage of Bacton terminal). Case study
Case 1
Case 2
Case 3
Case 4
Gas for power generation (mcm) Gas for other uses (mcm)
35.78 125.36
4.76 12.41
0.00 70.95
0.00 0.00
Table 6 Operational costs in £M (Outage of Bacton terminal). Case study Case Case Case Case
1 2 3 4
Electrical network
Gas network
Total
78.92 78.92 117.28 107.56
2543.30 995.91 1567.71 809.06
2622.22 1074.83 1684.99 916.62
4.2. Outage of Bacton gas terminal In this section, the complete outage of Bacton gas terminal was considered. This event would be triggered by the gas interruption through UK Continental Shelf (UKCS), BBL and IUK (see Fig. 4). In Fig. 14 the gas load shedding over the week in the gas network is presented. Due to the inertia of gas, as it was expected, close to the Bacton gas terminal largest load shedding was happened. In Table 5, it is shown that through Integrated there was no need to shed any gas demand for power generation as the security supply constraints of the gas network were taken into account. In Case 3, the modeling was done simultaneously and therefore the dispatch of the power generation had been modified in order to reduce gas for power generation, but there was not enough gas infrastructure flexibility to deliver gas to the demand centers. From Tables 4 and 6 it is concluded that value of gas infrastructure flexibility was increased with the severity of the contingency in the gas network. It is shown about 36% improvement in operational cost in Case 2 in respect to Case 3 in complete outage of Bacton gas terminal. Consequently, the contingency mode was more effectively managed in Case 2 rather to Case 3. As it was expected, the lowest operational cost was achieved through combining the two suggested solutions in this research (Case 4). In this mode, even in contingency conditions the gas and electricity networks were securely operating. The solutions helped to obviate the lack of gas supply with re-dispatching the generation mix in the power system and changing the gas flow direction in the gas network, which leaded to no gas demand curtailments. Through comparing the operation of the gas and electricity networks in different case studies it could be concluded that by taking into account the gas network infrastructure flexibility, the coupling link and interdependency of gas and electricity networks was decreased. The impact of a gas network incident on the electricity network was reduced and vice versa. Consequently, the security of the energy systems was increased. 5. Conclusion In order to improve the security of supply of gas and electricity networks, two solutions were presented. The first solution was to change the operation practice from sequential to an integrated approach to operate gas and electricity networks. The second solution was improving the flexibility of gas network infrastructure (multi-directional compressor stations). Through the modeled gas compressors, the gas flow could be redirected when it was required. Whereas, in one-directional compressor stations, the gas flow was dictated by the system operator. The system-wide
579
performance of these two solutions in dealing with wind intermittency was evaluated. It was shown that through an integrated operational practice significant amount of gas load shedding in contingency situations was avoided as the gas network security constraints were considered in the operation of the power system. In this research the contingency was related to partial and complete outage of the largest gas terminal in the GB gas network in 2030. Using flexible multidirectional compressor stations, reduced the total operational cost of the networks compared to one-directional compressor stations, due to the additional flexibility that was given to the network to deliver gas to the demand centers. When the severity of the gas network incident was increased the impact of gas network infrastructure flexibility was more noticeable. In another words, the coupling link between gas and electricity networks was decreased when flexibility was provided in the gas network infrastructure. The best operational management of the networks was achieved through combining the proposed solutions. Acknowledgments The authors gratefully acknowledge the Whole Systems Energy Modelling Consortium (wholeSEM) for providing funding for this work under EPSRC grant EP/K039326/1. References [1] UK Future Energy Scenarios 2014. National Grid plc.; Jul. 2014. [2] Fourth Carbon Budget Review-part 2. Committee on climate change; Dec. 2013. [3] Gas Ten Year Statement: UK Gas transmission. National Grid plc.; Dec. 2014. [4] Ma J, Silva V, Belhomme R, Kirschen DS, Ochoa LF. Evaluating and planning flexibility in sustainable power systems. IEEE Trans Sustain Energy 2013;4 (1):200–9. [5] Lu S, Makarov YV, Zhu Y, Lu N, Kumar NP, Chakrabarti BB. Unit commitment considering generation flexibility and environmental constraints. In: IEEE PES general meeting; 2010. p. 1–11. [6] Pudjianto D, Aunedi M, Djapic P, Strbac G. Whole-systems assessment of the value of energy storage in low-carbon electricity systems. IEEE Trans Smart Grid 2014;5(2):1098–109. [7] Troy N, Flynn D, O’Malley M. Multi-mode operation of combined-cycle gas turbines with increasing wind penetration. IEEE Trans Power Syst 2012;27 (1):484–92. [8] Mejia-Giraldo D, McCalley JD. Maximizing future flexibility in electric generation portfolios. IEEE Trans Power Syst 2014;29(1):279–88. [9] Liu C, Shahidehpour M, Fu Y, Li Z. Security-constrained unit commitment with natural gas transmission constraints. IEEE Trans Power Syst 2009;24 (3):1523–36. [10] Xydas E, Qadrdan M, Marmaras C, Cipcigan L, Jenkins N, Ameli H. Probabilistic wind power forecasting and its application in the scheduling of gas-fired generators. Appl Energy 2017;192:382–94. [11] Qiao Z, Guo Q, Sun Hongbin, Pan Z, Liu Y, Xiong W. An interval gas flow analysis in natural gas and electricity coupled networks considering the uncertainty of wind power. Appl Energy 2016. [12] Correa-Posada CM, Sanchez-Martin P. Integrated power and natural gas model for energy adequacy in short-term operation. IEEE Trans Power Syst 2015;30 (6):3347–55. [13] Zlotnik A, Roald L, Backhaus S, Chertkov M, Andersson G. Coordinated scheduling for interdependent electric power and natural gas infrastructures. IEEE Trans Power Syst 2017;32(1):600–10. [14] Zheng J, Wua Q, Jing Z. Coordinated scheduling strategy to optimize conflicting benefits for daily operation of integrated electricity and gas networks. Appl Energy 2017:370–81. [15] Gil M, Duenas P, Reneses J. Electricity and natural gas interdependency: comparison of two methodologies for coupling large market models within the european regulatory framework. IEEE Trans Power Syst 2016;31(1):361–9. [16] Chaudry M, Jenkins N, Qadrdan M, Wu J. Combined gas and electricity network expansion planning. Appl Energy 2014;113:1171–87. [17] Qadrdan M, Chaudry M, Wu J, Jenkins N, Ekanayake J. Impact of a large penetration of wind generation on the GB gas network. Energy Policy 2010;38 (10):5684–95. [18] Qadrdan M, Wu J, Jenkins N, Ekanayake J. Operating strategies for a gb integrated gas and electricity network considering the uncertainty in wind power forecasts. IEEE Trans Sustain Energy 2014;5(1):128–38. [19] Clegg S, Mancarella P. Integrated modeling and assessment of the operational impact of power-to-gas (P2G) on electrical and gas transmission networks. IEEE Trans Sustain Energy 2015;6(4):1–11.
580
H. Ameli et al. / Applied Energy 202 (2017) 571–580
[20] Qadrdan M, Ameli H, Strbac G, Jenkins N. Efficacy of options to address balancing challenges: integrated gas and electricity perspectives. Appl Energy 2017;190:181–90. [21] Üster H, Dilaveroglu A. Optimization for design and operation of natural gas transmission networks. Appl Energy 2014;133:56–69. [22] Sanaye S, Mahmoudimehr J. Optimal design of a natural gas transmission network layout. Chem Eng Res Des 2013;91(12):2465–76. [23] Trovik K, Fodstad M, Hellemo L. Optimization model to analyse optimal development of natural gas fields and infrastructure. Energy Procedia 2015;64 (1876):111–9.
[24] Skea J, Chaudry M, Wang X. The role of gas infrastructure in promoting uk energy security. Energy Policy 2012;43:202–13. [25] Spectra Energy. Changing gas flows to meet market needs.. [26] Electricity Ten Year Statement 2014. National Grid plc; Nov. 2014. [27] Chaudry M, Jenkins N, Strbac G. Multi-time period combined gas and electricity network optimisation. Electr Power Syst Res 2008;78(7):1265–79. [28] Gas Transportation Transmission Planning Code. National Grid plc; Oct. 2008. [29] Fico Xpress Optimisation Suite. .
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Value of gas network infrastructure flexibility in supporting cost effective operation of power systems Hossein Ameli a,⇑, Meysam Qadrdan b, Goran Strbac a a b
Control and Power Group, Imperial College, London, United Kingdom Institute of Energy, Cardiff University, Cardiff, United Kingdom
h i g h l i g h t s Integration of renewable energies causes operational challenges in power systems. Value of operational practices for gas and electricity networks were quantified. Role of gas network flexibility to address the challenges were evaluated. Less gas load curtailment when the gas network is more flexible. Cost savings when integrated approach and gas network flexibility were considered.
a r t i c l e
i n f o
Article history: Received 28 March 2017 Received in revised form 3 May 2017 Accepted 18 May 2017
Keywords: Integrated gas and electricity networks Wind generation variability Economic dispatch Flexibility Multi-directional compressors
a b s t r a c t The electricity system balancing is becoming increasingly challenging due to the integration of Renewable Energy Sources (RES). At the same time, the dependency of electricity network on gas supply system is expected to increase, as a result of employing flexible gas generators to support the electricity system balancing. Therefore the capability of the gas supply system to deliver gas to generators under a range of supply and demand scenarios is of a great importance. As potential solutions to improve security of gas and electricity supply, this paper investigates benefits of employing flexible multi-directional compressor stations as well as adopting a fully integrated approach to operate gas and electricity networks. A set of case studies for a GB gas and electricity networks in 2030 have been defined to quantify the value of an integrated operation paradigm versus sequential operation of gas and electricity networks. The results indicate there are significant overall system benefits (up to 65% in extreme cases) to be gained from integrated optimization of gas and electricity systems, emphasizing the important role of gas network infrastructure flexibility in efficiently accommodating the expected expansion of intermittent RES in future power systems. Ó 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http:// creativecommons.org/licenses/by/4.0/).
1. Introduction The share of Renewable Energy Resources (RES) is projected to increase significantly across the Great Britain (GB) network to meet the de-carbonization target in the future [1]. The installed capacity of wind power generation is expected to be between 52 GW [1] and 64 GW [2] in 2030 GB power system. This increase in wind generation share leads to balancing challenges due to the variable nature of the wind. In this context, flexible gas-fired generation with fast ramping rates is expected to facilitate cost effective integration of wind generation to meet the balancing challenges of ⇑ Corresponding author. E-mail addresses: [email protected] (H. Ameli), [email protected] (M. Qadrdan), [email protected] (G. Strbac).
supply and demand. The interdependency of the gas and electricity networks is increasing because of the role of gas-fired plants as the main linkage of these networks. Thus, the wind variability is reflected as intermittency in the gas demand as well. Based on higher inertia of gas, delivery time from supply sources (terminals and storage facilities) to demands in the gas network is much higher than electricity transmission. Hence, the Gas System Operator (GSO) maintains minimum volumes of gas, called Linepack, within the pipelines to meet rapid changes in the network. The variability of gas demand for power generation, consequently influences the linepack and makes its management more difficult. Real operational data from National Grid indicates that the linepack of the GB high pressure gas network in 2012 fluctuated with larger magnitude compared to year 2002 [3].
http://dx.doi.org/10.1016/j.apenergy.2017.05.132 0306-2619/Ó 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
572
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Nomenclature
Parameters & variables C cost (£) CR maximum compressor pressure ratio e produced green house gas emissions (tonnes) F objective function (£) N set P electrical power (MW) p gas pressure (Pa) p upper pressure bound (Pa) p lower pressure bound (Pa) Q volumetric gas flow rate (m3 =s) T time horizon (h) s time step (= 1 h) l compressor fuel rate coefficient (0.084 m3 =MJ) c polytropic efficiency (4.70 MJ=m3 ) g efficiency (%) f amount of gas trapped by a compressor (m3 =s) Superscripts a!b from node a to node b comp compressor dis discharge ecomp electrically-driven compressors elec electricity eload electrical demand em emission eshed electrical load shedding
Several studies have examined the role of flexibility options in addressing the balancing challenges driven by RES in power systems [4–8]. Flexibility is described as the ability of the system to deal with uncertainty and variability in the network, while keeping an acceptable level of reliability, from time scales of seconds to hours [4]. In [5] a framework considering generation flexibility requirements and environmental constraints for the Unit Commitment (UC) process is proposed. It was concluded, through utilizing flexible plants the occurrences of supply and demand imbalances caused by the wind forecast errors was decreased. In [7] it is shown that through using multi-mode operation of Combined Cycle Gas Turbines (CCGTs) the system security was increased. The multimode operation of CCGTs is related to the units which incorporate a steam bypass stack. These units are capable also to operate in open-cycle mode, which increase the flexibility of these gas plants in order to deal with large integration of wind into the system. Although less efficiency was achieved when operating in opencycle mode, however it was shown that in overall the emissions were significantly reduced as the generation portfolio has been changed and the production of generation units characterized by high emissions is decreased. The focus of these studies was mainly on the impact of the flexibility options in power systems operation, while the integrated energy systems value (gas and electricity) of flexibility has not been taken into account. Although, the interaction of gas and electricity networks have been studied from security of supply perspectives [9], wind forecasting and uncertainty [10,11], energy adequacy [12], coordinated scheduling [13,14], market coupling [15], and also planning strategy [16], however comprehensive analysis of large penetration of RES on the gas network operation has been reported by couple of studies [12,17–20]. The value of an integrated operation practice of the gas and electricity networks in comparison to sequential operation of these networks, has not
gload gas demand gshed gas load shedding int integrated pgen gas for power generation pumpinj injection to the pump storage unit pumpwith withdrawal from the pump storage unit sd shut-down storinj injection to the storage facility storwith withdrawal from the storage facility su start-up suc suction supp supply wavail available wind wcur curtailed wind var variable Subscripts b electrical busbar c compressor station g power generation unit k thermal power generation unit m gas node l gas pipeline p pump unit s gas storage facility t time y gas terminal
been evaluated in these studies. Furthermore, in the literature, electricity side flexibility options are discussed and to the best of our knowledge the value of gas network flexibility has not been investigated. In a few studies, including [21–23] the planning and operation optimization of the gas transmission networks from different aspects are studied. Additional pipelines and compressor stations are the main reinforcement components in the planning mode of these studies. In [24], an assessment on the role of gas network infrastructure in providing energy security has been studied. But, none of these studies considered the role of multi-directional compressor units in operation of the gas network. Although the role of multi-directional compressor as a mean of providing flexibility to the gas network is discussed in [25], however the research was focused on the gas network infrastructures and the value of this flexibility option was not quantified. As mentioned above, the future growth of wind generation capacity will increase the complexity of gas network management. One of the solutions to deal with these challenges is to make the gas network infrastructures flexible. Gas storage facilities and flexible multi-directional compressor stations are potential candidates to cope with the growing variabilities in the gas network. The interest of the GB system operator to invest in more flexible stations able to compress gas in different directions is reported in [26]. For example, if the available gas in North Sea could not be delivered to the network, one-directional compressor stations are not able to compensate for the lack of gas supply to meet the demand. Thus, redirecting the gas flow through the compressors could help to avoid additional reinforcement in the network. In this paper, potential solutions for improving capability of gas and electricity supply system to meet the demands were investigated. The solutions are (a) changing the operation practice through adopting an integrated approach to operate gas and electricity networks, (b) employing flexible multi-directional
573
H. Ameli et al. / Applied Energy 202 (2017) 571–580
compressor stations in the gas transmission network, and (c) the first two solutions combined. In the operation of gas and electricity networks, the multi-directional compressors are modeled so that the gas flow could be redirected to ensure that the gas demand is met while minimizing the operational cost of the gas and electricity networks. The value of multi-directional compressors were especially highlighted, when large variations in the gas network was happened (e.g., post-contingency conditions). Therefore, a set of gas supply interruption scenarios for a GB gas and electricity network in 2030 were defined. The efficacy of the above solutions to mitigate the impacts of the gas supply interruptions on the operation of gas and electricity networks were quantified. In these scenarios as a contingency was happened in the gas supply system, difficulties in the operation of the gas network was occurred. Due to the fact that location of majority of the gas plants were not very close to the gas supply points, gas might not be delivered in time due to limitations in speed in transporting gas through gas transmission grid. In order to compensate for limited electricity production by gas plant, more expensive options were required to produce more and imports from interconnection were increased. On the other hand, as the flexibility in the gas network was enhanced through multi-directional gas compressors, gas delivery to the gas plants was possible. Consequently, the overall emission and the total operational costs of the networks were reduced. 2. Modeling methodology 2.1. Integrated gas and electricity networks The enhanced version of the Combined Gas and Electricity Networks (CGEN) operation model [18] was used in order to represent a detailed model of flexible multi-directional compressor stations. This model was employed to examine the role of the gas network infrastructure flexibility in supporting future power systems with significant penetration of wind generation. The model is able to optimize the operation of the gas and electricity networks, simultaneously. Constraints of the power system modeled including characteristics of thermal generating units (i.e. ramp up/down, minimum up/down time, minimum stable generation of thermal units) and spinning reserve requirements, and energy balance (1) were considered. In the supply side of the energy balance, generation of the units, net energy of pump storage, absorbed wind by the grid are taken into account. On the other side (i.e., demand), electricity demand and electricity required for electrically-driven compressors are considered. The electrical load shedding term prevents the model from infeasibility, which shows the Value of Lost Electricity Load (VoLEL) in the electricity network. Eq. (1), indicates that at each time step (i.e., hourly) the supply and demand has to be balanced.
8t 2 T : ¼
8x 2 Nm ; t 2 T :
Q supp s þ Q x;t s þ ðQ storwith Q storinj Þs x;t x;t x;t
þ ðQ comp fx;t Þ s ¼ Q gload s þ Q pgen s Q gshed s x;t x;t x;t x;t
ð2Þ
Gas flow dynamics associated with linepack changes were captured in this granularity level. A quasi steady state condition for gas flow over a day was assumed. The intra-day variations of gas supply and demand were taken into possession in hourly time steps [27]. Detailed description and formulations of gas and electricity network operational modeling is presented in [20]. Gas-fired generating plants as well as electrically-driven compressors are the coupling components of electricity and gas networks. In electricity network, gas plants are electricity providers, while in gas network, gas-fired generating plants consume gas (i.e., gas load). On the other hand, electrically driven compressors consume electrical power (i.e., electrical load) and maintain the gas network pressure. An hourly time step rolling planning approach was applied for the network modeling. The main reasons for implementing a rolling planning approach was to enhance efficiency of computations by splitting into seven smaller-size optimization problems (from a week to a day duration). In the modeling, the electricity network states (e.g., ON/OFF state of the thermal generating units) of last hours of the previous day was taken into account for the next day. In the gas sector, based on the linepack and storage of the previous day the state of the gas network for the next day was updated. 2.1.1. Flexible multi-directional compressor stations To recover the loss of pressure in the gas infrastructure, which is caused by friction in the pipelines, installation of compressor units are considered. Compressors operate either electrical or gas driven. The compressor prime mover required power is presented in [27]. In order to model the multi-direction option of the compressor stations, the suction and discharge nodes of the compressor should be decision variables and not pre-set. Therefore, from the modeling point of view, it was assumed that at each compressor station node, virtually two compressor units are installed (from node a ! b, and b ! a (Fig. 1)), which compress the gas in opposite directions. At each time step only one of these compressors could operate. Through calculating the power consumption of these reverse compressors (3) and (4), the one with more power consumption (5) will dictate the direction of the gas flow at each time step. This means that the pressure in discharge node of the optimal compressor is more than the suction node. The pressure ratio of the discharge (outlet) node to suction (inlet) node is limited to CR (6). 8c 2 Nc ; t 2 T:
Ng Np Nb X X X Pg;t s þ ðP pumpwith Ppumpinj Þsþ Pwind s p;t p;t b;t g¼1
Nb X
ity energy balance, a gas load shedding term is considered in the gas flow balance which presents the Value of Lost Gas Load (VoLGL).
p¼1
ðPeload s Peshed s þ Pecomp sÞ b;t b;t b;t
b¼1
ð1Þ
b¼1
In the gas network modeling, constraints such as gas supply limits for terminals and storages, linepack changes, pressure limits, and gas compressor operation limits, and gas flow balance (2) were taken into account. The gas flow balance should be satisfied at each node and each time step. Gas provided through terminals, gas available in the pipelines, net gas from the storage facilities, and gas from the compressors excluding the gas consumed by gasdriven compressors are the key terms in the supply side. Two type of demands are considered; (a) required gas for the plants for power generation, and (b) gas for other uses. Same as the electric-
Fig. 1. A simplified structure of a flexible multi-directional compressor station.
574
H. Ameli et al. / Applied Energy 202 (2017) 571–580
2
Pa!b ¼ c;t
c:Q comp pbc;t c;t :4 a comp g pc;t 2
Pb!a ¼ c;t
c:Q comp pac;t c;t :4 b comp g pc;t
3
!ð1=cÞ
15
ð3Þ
3
!ð1=cÞ
15
ð4Þ
a!b Pcomp ¼ maxðPc;t ; Pb!a c;t c;t Þ
ðCRÞ
1
pac;t
6
pbc;t
6 CR
ð6Þ
Q comp 6 Q comp c;t c
ð7Þ
Pcomp c;t
ð8Þ
6
Pcomp c
pcomp c
ð9Þ
P pcomp c
ð10Þ
6
fc;t ¼ l:Pcomp c;t
ð11Þ
Flexible multi-directional compressor stations are able to flow the gas in different directions. For this purpose, it is required to reinforce the existing compressor units with additional piping and valves in the station yard. Currently, the GSO in GB is using Multi-junctions (i.e., pipelines, valves, and regulators) to control the flow of gas in the National Transmission System (NTS). Multijunctions can be located close to compressor stations, and therefore this make the gas compression in different directions possible [28]. Once the compressor becomes multi-directional, the gas demand will dictate the direction of the gas flow in the network [25]. This will enhance the gas network infrastructure flexibility, which improves the action of GSO for meeting the consumers requirements. A simplified structure of a flexible multidirectional compressor station including additional valves and pipelines are presented in Fig. 1. The red1(small dashed) line and the blue (dashed) line indicate the ability of modifying the gas flow in reverse directions. At each time the valves of the opposite direction are closed. The proposed model is able to change the direction of the gas flow in the compressors through comparing the pressures of different junctions in the station yards. 2.1.2. Objective functions In the electricity network, costs of power generation, electrical load shedding, and CO2 emission penalties were taken into account in the objective function (12). In the gas network the costs of gas supplies, storages, changes of the gas flow, and gas load shedding were considered (13). The objective function of the Integrated modeling was to minimize the costs of both networks (14).
F elec ¼
Ng Nk T X X X var sd em ðC fuel þ C Þ:P s þ ðC su g;t g g k þ C k þ C k :ek;t sÞ t¼1
g¼1
Nb X C eshed :Peshed s þ b;t
k¼1
!
ð12Þ
b¼1
F gas ¼
Ny Nl T X X X C gas :Q supp sþ C gas :@Q l;t s y;t t¼1
þ
y¼1
In Integrated approach, in order to avoid twice calculation of gas plants generation cost, C fuel of gas power plants are zero, as the gas consumption of these plants were considered as a type of demand in the gas network operation.
2.2. Integrated vs. Sequential strategy 2.2.1. Sequential modeling In the Sequential mode, the operation of electricity network (including UC/Economic Dispatch and power flow) was optimized and the gas demand for power generation was then calculated. In this approach, the gas demand for power generation was provided as an input to the gas network operation model. In this modeling, the operational costs of the electricity network were minimized (12) and afterwards the operational costs of the gas network were minimized (13). The algorithm of the Sequential modeling using a rolling planning approach is presented in Fig. 2. The red (dashed) line indicates the sequential link of the electricity and gas networks. The Sequential approach does not consider the gas supply constraints while optimizing operation of the power system. Hence, a key disadvantage of this modeling is that potential interruptions in the gas supply do not influence the dispatch of gas-fired plants in the electricity network, as the modeling of electricity system is carried out before the gas network.
2.2.2. Integrated modeling Unlike the Sequential mode, in the Integrated modeling the gas and electricity network constraints were considered synchronously. Fig. 3 illustrates the algorithm of the Integrated modeling. The Integrated model minimizes the total operational cost of the gas and electricity networks, simultaneously (14). The corresponding optimization problem of Integrated model was solved through Mixed Integer Non-Linear Programming (MINLP) approach in Xpress-IVE (i.e., mmnlp module) [29]. The total number of variables for the week is about 260 k variables including about 20 k integer variables. However, as mentioned previously, a rolling approach has been implemented and the optimization is divided into seven smaller optimization problems. The main reason of using a rolling approach is that due to the non-linearity as well as integer space of the model, by increasing the number of variables, the computational time would be increased exponentially. In order to get more accuracy for modeling each day, four hours of the next day is also considered. As a result, at each step of the rolling approach around 43 k variables including around 3500 integer variables has been optimized. A branch and bound strategy has been used to find the solution of the binary variables. The search has been stopped when the absolute tolerance of the best solution’s objective function and the current best solution bound is less than a set value which is presented in (15).
jF BestBoundj 6 0:1% maxðF; BestBoundÞ
l¼1
Ns Nm X X ðC storwith :Q storwith C storinj :Q storinj Þsþ C gshed :Q gshed s s;t s;t m;t s¼1
ð14Þ
ð5Þ
Each compressor was subjected to maximum flow rate, power consumption, and capacity constraints (7)–(10). The tapped gas was calculated through (11). 8c 2 Nc ; t 2 T:
pdis c;t psuc c;t
F int ¼ F elec þ F gas
!
m¼1
ð13Þ 1 For interpretation of color in Figs. 1 and 2, the reader is referred to the web version of this article.
ð15Þ
The model was performed on a 3.20 GHz Intel(R) Xeon(R) processor and 16 GB of RAM system. The simulation time for different case studies was between 93 and 118 min. Initial values based on output of the optimization for a single time step were given to all decision variables. In addition, the decision variables were scaled so that the order magnitude of the values do not vary significantly. These two procedures aided to obtain convergence and stability of the optimization process.
575
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Fig. 2. Sequential mode algorithm.
Base (Case 1): In this case the Sequential modeling of gas and electricity networks (see Fig. 2) was considered. Direction of gas flow through compressors (i.e. suction and discharge nodes of compressors) were pre-set. Flexible multi-directional compressor stations (Case 2): The compressors were assumed to have the capability to change the direction of gas flow (see Fig. 1). The suction and discharge nodes of each compressor were decision variables that were determined as part of the optimization problem of the gas network operation. Integrated operational practice (Case 3): In this case the operation of gas and electricity networks were optimized simultaneously (see Fig. 3). The directions of gas flow through compressors were pre-set. Fully flexible (Case 4): Both solutions presented in Case 2 and Case 3 were combined in this case study.
Fig. 3. Integrated mode algorithm.
Impacts of partial and complete outage of Bacton gas terminal, the largest gas terminal in GB, on the operation of the gas and electricity networks were investigated for different cases. Two outage scenarios were defined: (a) interruption of gas import from Balgzand to Bacton Line (BBL) and Interconnectors UK (IUK) and (b) interruption of gas supply from whole Bacton terminal (see, Fig. 4). These studies were to assess how Integrated approach along with flexible multi-directional compressor stations could mitigate the adverse effect of an incident in the gas network.
3. Case studies 3.1. Descriptions of the cases As it was mentioned previously, three solutions were suggested in this paper. Changing the operation practice (i.e., integrated approach of gas and electricity modeling), reinforcing the physical assets of the gas network (i.e., flexible multi-directional compressor stations), and combination of these two. To evaluate the capability of the solutions in improving the security of supply, following cases for the GB electricity and gas networks were examined:
Fig. 4. Bacton supply sources.
576
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Fig. 5. GB 29-Busbar electricity transmission network. Fig. 6. GB simplified NTS gas network.
3.2. GB gas and electricity 2030 networks A GB electricity transmission network including 29 busbars, 47 transmission lines and 148 generators based on [26] was implemented in this research (Fig. 5). An updated version of the simplified NTS network (63 nodes, 54 gas pipelines, 9 terminals, and 5 storages), introduced in [17] based on [3] was applied in the model (Fig. 6). 3.3. Input data
Table 1 Generation mix in GB 2030 [1]. Technology
Capacity (GW)
Wind Gas Interconnector Nuclear Coal with CCS Pumped storage Hydro Other
52 33 11.5 9 4.5 2.7 1.1 1.2
The GB gas and electricity networks were modeled over a typical winter week in 2030. Table 1 presents the power generation mix in this study (see Fig. 7). In the National Grid outlook in 2030, 52 GW wind generation is projected to be installed across GB [1]. The hourly wind generation, electricity demand data as well as non-electric gas demand data in 2030 were taken from [20]. It is worth noting that in the model wind generation data was defined as available wind power. The available wind power could be absorbed by the electricity system or had to be curtailed due to the excess of wind generation (16).
8b 2 Nb ; t 2 T :
Pwavail ¼ P wind þ Pwcur b;t b;t b;t
ð16Þ
Fig. 7. Hourly available wind power, gas demand, and electricity demand [20].
577
H. Ameli et al. / Applied Energy 202 (2017) 571–580
4. Results and discussion 4.1. Partial outage of Bacton gas terminal 4.1.1. Generation mix The operation of the electricity network in the Sequential cases (Case 1 and Case 2) were the same, as it was optimized irrespective to the operation of the gas network. In Integrated Cases (Case 3 and Case 4) the constraints of gas supply were taken into account. Changes in electricity generation by different technologies in Integrated cases compared to Sequential cases are presented in Fig. 8. The generation of nuclear and hydro did not vary considerably as these are lowest cost generation technologies [17]. Moreover, hydro capacity was relatively small (see Table 1), so these plants did not make significant contribution to meeting electricity demand. Total generation of gas plants over the week in Integrated cases was less than Sequential cases. The reason is that in the Integrated modeling the gas and electricity networks operation were optimized simultaneously, therefore gas supply constraints were considered in the economic dispatch of power generation. As a result, roughly 18,000 and 7,000 tonnes of CO2 production were reduced in Case 3 and Case 4 (compared to Sequential cases), respectively. Due to the flexible multi-directional compressor stations option, gas flow was redirected in different time steps in order to deliver gas from different supply points. Therefore, because of the provided flexibility in the gas network the gas plants in Case 4 generated about 40 (GWh) more than Case 3 over the time horizon. In Case 3 in order to meet the energy balance, about 35 (GWh) electrical energy was imported through interconnectors. Whereas, in Case 4, the power flow through interconnectors has been decreased. In Fig. 9-a, the role of gas plants as an economical option [17] able to complement the lack of wind generation during low-wind, high-demand period (105–118 h) is presented. In Sequential cases, the gas supply constraints were not considered in operation of the electricity network. Therefore, as the electricity network operation was modeled first (see Fig. 2), it was preferred operating gas plants rather to importing energy via interconnection to keep the energy balance in the network, as this minimizes the operation costs. On the other side in Integrated cases as gas supply interruption at Bacton terminal was taken into account in the operation of power system, the generation through gas plants was reduced compared to the Sequential cases. This was mostly due to the fact that less gas was available close to Bacton terminal, which has affected especially the gas plants in busbars 20 and 26. In Case 4, thanks to the gas flow redirection option, delivering gas to the aforementioned plants could be managed cost effectively. The reduced generation from gas plants in Case 3 and Case 4 were compensated with interconnectors. Fig. 9-b presents the contribution of interconnectors in providing the energy balance in low-wind, high-demand period. It is worth noting that although the price of importing energy via interconnectors was assumed to be higher
Fig. 8. Electricity generation changes in respect to Sequential cases.
Fig. 9. Power supplied during low-wind, high-demand periods: (a) Gas-plants, (b) interconnectors.
than gas plants [17], the interconnection prevented gas load shedding in the gas network. Whereas, in the Sequential cases the model selected seemingly lower cost option based on using gas plants rather interconnector, as the real-time gas availability was not considered.
4.1.2. Gas supply and linepack Total gas supply through terminals and gas storage facilities are presented in Table 2. Changing the operational practice as well as the provided flexibility improved the gas delivery to the demand centers by reducing the gas transmission congestion. Fig. 10 shows the linepack in the pipelines during the simulation horizon. The linepack on a particular day of the week changes based on the demand pattern for power generation on that day. The linepack fluctuations was related to the fact that in off-peak times (e.g. in the mornings before 06:00) the gas demand was low, and therefore linepack was increasing. In the peak-times
Table 2 Total gas supply over the week. Case study
Case 1
Case 2
Case 3
Case 4
Gas supply (mcm)
2285.8
2322.5
2311.6
2319.1
Fig. 10. Linepack in different cases.
578
H. Ameli et al. / Applied Energy 202 (2017) 571–580
Table 3 Total gas load shedding (Bacton partial outage). Case study
Case 1
Case 2
Case 3
Case 4
Gas for power generation (mcm) Gas for other uses (mcm)
5.62 29.69
0.85 2.27
0.00 0.00
0.00 0.00
(e.g. 17:00–20:00) the stored gas in the pipelines was used and the linepack has been reduced. The analysis of the amount of the linepack indicates more availability in the within-pipeline storage capability of the gas network in multi-directional cases (Case 2 and Case 4). This enables the gas network operators to manage rapid changes in the gas demand in short term. The gas load shedding in different cases is shown in Table 3. In Case 1 the total amount of gas load shedding throughout the week was about 35 million cubic meter (mcm) due to the low flexibility of the network. As explained before, in the Sequential cases, first the electricity network was modeled, and different states of the gas network were not considered. This means that the electricity network requested gas for generation without any constraints on gas supply. On the other side, as an outage occurred, the gas could not be delivered to the demands. In Case 2 there was flexibility in the gas network physical assets but the requested gas for power generation was still more than supply. Hence, in the modeling’s point of view about 0.85 (mcm) and 2.27 (mcm) of the gas for power generation and non-electric gas demand had to be curtailed, respectively. In Integrated modeling the above mentioned problem was obviated as the gas supply constraints were considered and so it was not required to shed any gas demand. A comparison between Case 1 and Case 3 indicates the importance of considering gas and electricity constraints synchronized as through proper scheduling of the gas plants there was no need to curtail gas demands. In the Sequential modeling, gas for power generation and other uses have the same value, and therefore there was no priority in curtailing different type of demands.
balancing issues. A positive power consumption means that the compressor was operating in the pre-set direction, whereas a negative value of power consumption indicates that the gas flow was redirected. In normal condition, the direction of the gas through compressors close to the gas terminals was same as the injected gas through terminals. Thus, when an incident in the gas terminals happened, compressor units close to the terminal stations had to modify the direction of the gas flow in order to meet the gas demand in areas close to the terminals and keep the gas pressure in the acceptable range as much as it was feasible. As an example, compressor station number nine (located in node 62, 63) mostly operated in the reverse direction. In addition, the power consumption by the compressors in Case 4 was less than Case 2 which means the gas network operation was managed better as it was less need to compress the pressure. 4.1.4. Gas and electricity networks costs Operational cost of the networks for various case studies over the typical winter week in partial outage of Bacton gas terminal are presented in Table 4. It is noteworthy that the penalty price of gas load shedding was assumed to be 11.1 £M=mcm [27]. The cost savings in different case studies are presented in Fig. 13. It is demonstrated that through operating simultaneously the gas and electricity networks and employing flexible multidirectional compressor stations, the lack of gas supply was completely compensated and about 30% cost saving in comparison to Case 1 was achieved.
Table 4 Operational costs in £M (Bacton partial outage). Case study Case Case Case Case
1 2 3 4
Electrical network
Gas network
Total
78.92 78.92 92.90 80.21
1190.66 834.49 808.18 810.39
1269.58 913.41 901.08 890.60
4.1.3. Compressor direction modification The proposed model was capable of changing the direction of the compressors. Mostly the compressors close to the Bacton gas terminal operated in reverse direction than the pre-set direction. In Figs. 11 and 12 it is shown that how flexible multi-directional compressor stations could provide flexibility to manage better the
Fig. 13. Cost savings in comparison to Case 1.
Fig. 11. Compressors power consumption in Case 2.
Fig. 12. Compressors power consumption in Case 4.
Fig. 14. Gas load shedding (Locations).
H. Ameli et al. / Applied Energy 202 (2017) 571–580 Table 5 Gas load shedding (Outage of Bacton terminal). Case study
Case 1
Case 2
Case 3
Case 4
Gas for power generation (mcm) Gas for other uses (mcm)
35.78 125.36
4.76 12.41
0.00 70.95
0.00 0.00
Table 6 Operational costs in £M (Outage of Bacton terminal). Case study Case Case Case Case
1 2 3 4
Electrical network
Gas network
Total
78.92 78.92 117.28 107.56
2543.30 995.91 1567.71 809.06
2622.22 1074.83 1684.99 916.62
4.2. Outage of Bacton gas terminal In this section, the complete outage of Bacton gas terminal was considered. This event would be triggered by the gas interruption through UK Continental Shelf (UKCS), BBL and IUK (see Fig. 4). In Fig. 14 the gas load shedding over the week in the gas network is presented. Due to the inertia of gas, as it was expected, close to the Bacton gas terminal largest load shedding was happened. In Table 5, it is shown that through Integrated there was no need to shed any gas demand for power generation as the security supply constraints of the gas network were taken into account. In Case 3, the modeling was done simultaneously and therefore the dispatch of the power generation had been modified in order to reduce gas for power generation, but there was not enough gas infrastructure flexibility to deliver gas to the demand centers. From Tables 4 and 6 it is concluded that value of gas infrastructure flexibility was increased with the severity of the contingency in the gas network. It is shown about 36% improvement in operational cost in Case 2 in respect to Case 3 in complete outage of Bacton gas terminal. Consequently, the contingency mode was more effectively managed in Case 2 rather to Case 3. As it was expected, the lowest operational cost was achieved through combining the two suggested solutions in this research (Case 4). In this mode, even in contingency conditions the gas and electricity networks were securely operating. The solutions helped to obviate the lack of gas supply with re-dispatching the generation mix in the power system and changing the gas flow direction in the gas network, which leaded to no gas demand curtailments. Through comparing the operation of the gas and electricity networks in different case studies it could be concluded that by taking into account the gas network infrastructure flexibility, the coupling link and interdependency of gas and electricity networks was decreased. The impact of a gas network incident on the electricity network was reduced and vice versa. Consequently, the security of the energy systems was increased. 5. Conclusion In order to improve the security of supply of gas and electricity networks, two solutions were presented. The first solution was to change the operation practice from sequential to an integrated approach to operate gas and electricity networks. The second solution was improving the flexibility of gas network infrastructure (multi-directional compressor stations). Through the modeled gas compressors, the gas flow could be redirected when it was required. Whereas, in one-directional compressor stations, the gas flow was dictated by the system operator. The system-wide
579
performance of these two solutions in dealing with wind intermittency was evaluated. It was shown that through an integrated operational practice significant amount of gas load shedding in contingency situations was avoided as the gas network security constraints were considered in the operation of the power system. In this research the contingency was related to partial and complete outage of the largest gas terminal in the GB gas network in 2030. Using flexible multidirectional compressor stations, reduced the total operational cost of the networks compared to one-directional compressor stations, due to the additional flexibility that was given to the network to deliver gas to the demand centers. When the severity of the gas network incident was increased the impact of gas network infrastructure flexibility was more noticeable. In another words, the coupling link between gas and electricity networks was decreased when flexibility was provided in the gas network infrastructure. The best operational management of the networks was achieved through combining the proposed solutions. Acknowledgments The authors gratefully acknowledge the Whole Systems Energy Modelling Consortium (wholeSEM) for providing funding for this work under EPSRC grant EP/K039326/1. References [1] UK Future Energy Scenarios 2014. National Grid plc.; Jul. 2014. [2] Fourth Carbon Budget Review-part 2. Committee on climate change; Dec. 2013. [3] Gas Ten Year Statement: UK Gas transmission. National Grid plc.; Dec. 2014. [4] Ma J, Silva V, Belhomme R, Kirschen DS, Ochoa LF. Evaluating and planning flexibility in sustainable power systems. IEEE Trans Sustain Energy 2013;4 (1):200–9. [5] Lu S, Makarov YV, Zhu Y, Lu N, Kumar NP, Chakrabarti BB. Unit commitment considering generation flexibility and environmental constraints. In: IEEE PES general meeting; 2010. p. 1–11. [6] Pudjianto D, Aunedi M, Djapic P, Strbac G. Whole-systems assessment of the value of energy storage in low-carbon electricity systems. IEEE Trans Smart Grid 2014;5(2):1098–109. [7] Troy N, Flynn D, O’Malley M. Multi-mode operation of combined-cycle gas turbines with increasing wind penetration. IEEE Trans Power Syst 2012;27 (1):484–92. [8] Mejia-Giraldo D, McCalley JD. Maximizing future flexibility in electric generation portfolios. IEEE Trans Power Syst 2014;29(1):279–88. [9] Liu C, Shahidehpour M, Fu Y, Li Z. Security-constrained unit commitment with natural gas transmission constraints. IEEE Trans Power Syst 2009;24 (3):1523–36. [10] Xydas E, Qadrdan M, Marmaras C, Cipcigan L, Jenkins N, Ameli H. Probabilistic wind power forecasting and its application in the scheduling of gas-fired generators. Appl Energy 2017;192:382–94. [11] Qiao Z, Guo Q, Sun Hongbin, Pan Z, Liu Y, Xiong W. An interval gas flow analysis in natural gas and electricity coupled networks considering the uncertainty of wind power. Appl Energy 2016. [12] Correa-Posada CM, Sanchez-Martin P. Integrated power and natural gas model for energy adequacy in short-term operation. IEEE Trans Power Syst 2015;30 (6):3347–55. [13] Zlotnik A, Roald L, Backhaus S, Chertkov M, Andersson G. Coordinated scheduling for interdependent electric power and natural gas infrastructures. IEEE Trans Power Syst 2017;32(1):600–10. [14] Zheng J, Wua Q, Jing Z. Coordinated scheduling strategy to optimize conflicting benefits for daily operation of integrated electricity and gas networks. Appl Energy 2017:370–81. [15] Gil M, Duenas P, Reneses J. Electricity and natural gas interdependency: comparison of two methodologies for coupling large market models within the european regulatory framework. IEEE Trans Power Syst 2016;31(1):361–9. [16] Chaudry M, Jenkins N, Qadrdan M, Wu J. Combined gas and electricity network expansion planning. Appl Energy 2014;113:1171–87. [17] Qadrdan M, Chaudry M, Wu J, Jenkins N, Ekanayake J. Impact of a large penetration of wind generation on the GB gas network. Energy Policy 2010;38 (10):5684–95. [18] Qadrdan M, Wu J, Jenkins N, Ekanayake J. Operating strategies for a gb integrated gas and electricity network considering the uncertainty in wind power forecasts. IEEE Trans Sustain Energy 2014;5(1):128–38. [19] Clegg S, Mancarella P. Integrated modeling and assessment of the operational impact of power-to-gas (P2G) on electrical and gas transmission networks. IEEE Trans Sustain Energy 2015;6(4):1–11.
580
H. Ameli et al. / Applied Energy 202 (2017) 571–580
[20] Qadrdan M, Ameli H, Strbac G, Jenkins N. Efficacy of options to address balancing challenges: integrated gas and electricity perspectives. Appl Energy 2017;190:181–90. [21] Üster H, Dilaveroglu A. Optimization for design and operation of natural gas transmission networks. Appl Energy 2014;133:56–69. [22] Sanaye S, Mahmoudimehr J. Optimal design of a natural gas transmission network layout. Chem Eng Res Des 2013;91(12):2465–76. [23] Trovik K, Fodstad M, Hellemo L. Optimization model to analyse optimal development of natural gas fields and infrastructure. Energy Procedia 2015;64 (1876):111–9.
[24] Skea J, Chaudry M, Wang X. The role of gas infrastructure in promoting uk energy security. Energy Policy 2012;43:202–13. [25] Spectra Energy. Changing gas flows to meet market needs.