Dynamic analysis of hybrid energy systems under flexible operation and variable renewable generation – Part I: Dynamic performance analysis

Energy 52 (2013) 1e16

Contents lists available at SciVerse ScienceDirect

Energy journal homepage: www.elsevier.com/locate/energy

Dynamic analysis of hybrid energy systems under flexible operation and variable renewable generation e Part I: Dynamic performance analysis Humberto E. Garcia*, Amit Mohanty, Wen-Chiao Lin, Robert S. Cherry Idaho National Laboratory, 2525 N. Fremont Drive, Idaho Falls, ID, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 January 2012 Received in revised form 25 October 2012 Accepted 23 November 2012 Available online 19 March 2013

Dynamic analysis of HES (hybrid energy systems) under flexible operation and variable renewable generation is considered in this two-part communication to better understand various challenges and opportunities associated with the high variability arising from integrating renewable energy into the power grid. Unique consequences are addressed by devising advanced HES solutions in which multiple forms of energy commodities, such as electricity and chemical products, may be exchanged. Dynamic models of various unit operations are developed and integrated within two different HES options. One HES option, termed traditional, produces electricity only and consists of a primary heat generator, a steam turbine generator, a wind farm, and a battery storage. The other HES option, termed advanced, includes not only the components present in the traditional option but also a chemical plant complex to repurpose excess energy for non-electricity services, such as for the production of chemical goods. In either case, a given HES is connected to the power grid at a point of common coupling and requested to deliver a certain electricity generation profile as dictated by a regional power grid operator based on a predicted demand curve. A dynamic performance analysis of these highly-coupled HES is conducted in this part one of the communication to identify their key dynamical properties and limitations and to prescribe solutions for best managing and mitigating the high variability introduced from incorporating renewable energy into the energy mix. Ó 2013 Published by Elsevier Ltd.

Keywords: Hybrid energy systems Energy system dynamic analysis Flexible operation Variable renewable generation

1. Introduction 1.1. Background Energy utilization in the U.S., rather than exhibiting a diverse landscape of energy flows that provide energy security by not having any one energy use overly reliant on just one energy source, is instead has rather concentrated source-use pairings [1]. Each major natural energy source is primarily used for one purpose: nuclear and coal for electricity, natural gas for heating (with a modest fraction going to electricity), and petroleum for transportation fuels. Each purpose is likewise dependent on one or only a few sources: transportation on petroleum, heating on natural gas and electricity, and electricity primarily on coal but with significant fractions coming from nuclear and natural gas. Other sources like geothermal, solar, wind, and biomass make only very minor contributions. This constricted architecture may lead into undesired * Corresponding author. Tel.: þ1 208 526 7769; fax: þ1 208 526 3677. E-mail address: [email protected] (H.E. Garcia). 0360-5442/$ e see front matter Ó 2013 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.energy.2013.01.022

consequences or externalities if any one of these sources is disrupted as happened in the 1970s oil embargoes or with the longterm shutdown of nuclear reactors in Japan and elsewhere after the Fukushima accident. The upsets might also be in the uses of energy, for instance if fuel cell vehicles or plug-in hybrid vehicles running primarily on electricity become predominant in the next 20e40 years. The abandonment of coal for space heating over the last century is another example. The consequences of these types of events can be social, economic, geopolitical, or environmental in nature. To provide more robustness to the U.S.’s and the world’s energy supply network, a more flexible energy flow landscape should be developed that could, for example, to a greater extent use NG (natural gas), biomass, coal, and nuclear energy for the production of transportation fuels. This leads to the definition of a hybrid energy system: multiple energy inputs converted to multiple energy products using interacting complementary conversion processes. Liquid transportation fuels similar to existing petroleum-derived fuels can efficiently be produced from a carbon source and a separate energy source (Table 1).

2

H.E. Garcia et al. / Energy 52 (2013) 1e16

Nomenclature AEG AG AGD AHG CAES CP CPC EA ESE GHG HES HTE MIMO

MISO NG ISO I&C O&M PCC PHG PP REN RG RP RTO SMR STG USD

auxiliary electricity generation achievable generation average generation deficit auxiliary heat generation compressed air energy storage chemical plant chemical plant complex energy accommodation energy storage element greenhouse gas hybrid energy system(s) heat transfer element multiple input, multiple output

By adding non-traditional energy sources, such as renewable generation, and products, such as transportation fuels, energy system hybridization is a promising strategy to achieve energy security through diversification and integration of energy portfolios. In this manner, not only undesirable economic conditions but also environmental concerns can be resolved, as fuel transportation solutions using alternative fuels (e.g., biomass to liquids with carbon capture and storage) often produce less GHG (greenhouse gas) emissions than using conventional fuels. In order to reduce pollution and dependency on oil, a coordinated energy strategy may aim to derive electricity from clean-energy sources (e.g., nuclear and renewable energy) and to produce transportation fuels from regional carbon resources (e.g., NG, coal, and biomass). Higher levels of renewable penetration in the current energy portfolio are consequently a desirable goal as a means of attaining improved resource utilization and environmental sustainability. As a decentralized power generation alternative, often called microgrids, HES (Hybrid energy systems) can utilize available energy resources in an efficient and cost-effective manner, as suggested in Refs. [2,3,4]. Advanced HES configurations discussed in this paper aim to:  facilitate effective integration of renewable energies;  promote usage of alternative carbon sources (e.g., NG, biomass, coal) for production of chemical products (such as transportation fuel);  reduce environmental impact;  enhance both power (electricity) and energy management, in addition to reliability and security;  support smooth integration of diverse energy sources and products within existing power and fueling infrastructures. Given the above drivers, HES that can accommodate high renewable penetration are increasingly receiving substantial consideration. Wind generation has grown tremendously in the last

Table 1 Possible feeds and products of a hybrid energy system. Energy sources

Carbon sources

Energy products

Nuclear Wind Solar thermal Solar photovoltaic Geothermal Combustion or oxidation of carbon (but emits CO2)

Natural gas Coal Biomass CO2 captured from flue gas CO2 captured from atmosphere

Electricity Gasoline Diesel fuel Hydrogen Commodity chemicals such as methanol, ethylene, or ammonia District heat

multiple input, single output natural gas independent system operator instrumentation & controls operations & maintenance point of common coupling primary heat generation power plant renewable generation required generation renewable penetration Regional Transmission Organization small modular reactor steam turbine generator US dollar

decade and Renewable Portfolio Standards will assure that wind energy continues to be significant. This work emphasizes the dynamic analysis of HES under flexible operation and variable renewable generation, without considering how federal and state policies might affect the manner in which future electric power is generated and controlled in the U.S. Thus, the paper assumes that renewable energy is generated on a must-take basis irrespective of the economics and the logistics of how it is generated, sold, transmitted, bought, and used. Given this must-take assumption, any other energy alternative that may be more profitable, environment friendly (lower life-cycle CO2 emission), or user-friendly as compared to renewable are not explored in this work. 1.2. State of the art Neither the concept of combining various sources of energy nor the multiple utilization of energy produced is fundamentally new. Numerous researchers as well as academicians have explored these ideas to various degrees. Usually, the literature available suggests HES to act as energy supplier in three different contexts e first, as a grid-energy supplier; second, as supplier of electricity in off-thegrid locations; and third, as provider for production of energy products (e.g., hydrogen, fresh water, transportation fuel). There are many examples of HES being proposed to act in a stand-alone manner at off-the-grid locations. In Ref. [5], authors applied total sites targeting, a successful strategy used in process industries to locally integrated energy sectors. This investigation showed that by using methods, such as total sites targeting, renewable can be integrated to energy mix even at a local level. Dali et al. showed HES capacity to operate in grid-connected as well as stand-alone mode under suitable control and energy management [6]. In their experimental work, the HES included wind and PV (photovoltaic) physical emulators, battery energy storage, load and a controlled interconnection to the LV (low voltage) grid. In Ref. [7], the cost-effectiveness of the solar PV system and the solar/hydro schemes for rural electrification, which are considerably different from conventional electric grid, are evaluated and shown to be more reliable and sustainable than the use of a diesel genset. Similarly, hybrid diesel power plants with high-penetration renewable and compressed air energy storage were explored for off-grid rural electrification in Ref. [8]. In Ref. [9], a technoeconomic feasibility study of hybrid PV/diesel HES system over diesel gensets in a Malaysian remote area was conducted. The study showed that the fluctuating price and GHG emission of diesel fuel can be countered to great extent by an HES utilizing renewable sources. Similar case-studies were conducted by Reichling et al. for

H.E. Garcia et al. / Energy 52 (2013) 1e16

Minnesota, US in Ref. [10], by Erdil et al. for Cyprus in Ref. [11], by Rehman et al. for a Saudi Arabian village in Ref. [12] and by Nandi et al. for a Bangladeshi village in Ref. [13]. Performance evaluation of an HES utilizing high temperature fuel cells along with a state-ofart radial turbo machinery (e.g., gas turbines) was conducted in Ref. [14]. This performance evaluation covered three aspects of analysis, namely, on-design, off-design, and control. It showed that pressurized fuel-cell based HES are superior to non-pressurized ones under on-design and off-design conditions. However, it was also concluded that first generation of these types of HES may not be able to respond to load variations; thus, prohibiting use of renewable in the energy pool. In Ref. [15], investigation was carried out to provide Ireland a clean sustainable energy source by introducing an HES consisting of wind and hydrogen technology. In Ref. [16], dynamic simulation of PV/Wind HES was conducted for a stand-alone mini-grid situation consisting of different wind turbine and PV array models. It was also recommended to add special devices for electricity storage, so that constant electricity output over the whole year may be achieved. HES has also found significant traction for non-power related energy use. In Ref. [17], the possibility of using a fleet of plug-in hybrid vehicles was considered to avoid mismatch in future demand and supply when high-penetration renewable are put into the energy mix along with less flexible power generation units such as nuclear power plants. In Refs. [18,19], a large scale evaluation of HES has been conducted for powering reverse osmosis desalination unit and providing fresh water. In Ref. [20], the use of HES was proposed for sustainable desert agriculture in Egypt to increase total food production under desert land reclamation program. The power, process, and other related industries are paying increasingly more attention to how they generate, store and use energies using HES. This shows that the HES have made transition from a conceptual discourse in academic setting to commerciallyfeasible industrial and military applications. For example, the U.S. Army is considering the development of mobile and stationary HES with various traditional and renewable energy sources [21e24]. There are also several examples of existing or proposed commercial use of HES, finding wide acceptance from computer to power industries. Companies like IBM and HP are proposing and implementing several unique HES concepts [25]. Florida Power and Light Company is working on novel power plant concepts utilizing solar and clean coal technologies [26,27]. Xcel energy company is experimenting with parabolic-trough solar technology integrated with a coal-fired power plant at its Cameo Generating Station [28]. GE introduced its new generation combined cycle power plant, the FlexEfficiency 50 power plant, rated 510 MW of power output, which uses NG (natural gas) as its primary energy sources to provide backup power for intermittent sources of clean electricity, such as solar or wind power plants [29]. Although expected to provide important benefits, it has been largely recognized by industry and academia that increasing renewable penetration poses great technical challenges in terms of grid integration and stability [30e33]. This is due to the unpredictability, nondispatchability, and high variability associated with renewable energy sources, such as wind and solar power. Some renewable generation sources (e.g., wind, solar) depend upon resource availability, while others (e.g., hydroelectric and geothermal) have various degrees of dispatchability depending on physical, economic, and environmental constraints. For example, it has been noted that wind generation presents variability on every time scale [34]. Further complicating its integration, renewable generation is typically dispatched as “must-take”, with their production trends usually out of phase with demand [35]. While small levels of renewable penetration have tolerable effects on grid operation, high levels require significant changes to

3

traditional energy systems topology and grid infrastructure architecture. If this variability is not mitigated by engineering solutions, it has to be accommodated by the grid, which has often been found to be prohibitive. In general, it may be more cost-effective and less complex to attenuate the variability introduced by renewable energy before feeding their electricity contributions into the power grid, as opposed to modifying the existing power grid in order to handle renewable variability. This attenuation may be accomplished by using energy storage devices such as batteries and energy compensation (e.g., [30]). Methods like improving LVRT (lowvoltage ride through) of wind turbines using novel power electronics circuits [36,37], flexible fossil fuel plant, diesel stand-by, demand flexibility [38], and variable speed wind turbines [39] are not sufficiently capable to immunize the power grid from renewable variability. However, it is argued here that a more effective strategy is to extend the architecture of traditional HES to enable multiple energy commodity exchanges, including not only dispatchable electricity but also other energy storing products, such as methanol, ammonia, syngas, and hydrogen. HES considered here exhibit well-defined physical boundaries, functioning as energy centers consisting of multiple closelycoupled energy components. Fig. 1 illustrates the energy grid topology assumed for incorporating HES into the power grid. In this energy grid topology, there are broadly three different types of units: traditional generation units (Gi), hybrid energy systems (HESi), and loads (Li). Traditional generation units include not only large electric power plants (PP) (e.g., nuclear PP, hydro PP, and fossil-fuel-based PP) but also co-fired plants (e.g, coal-fired PP that also burns a small fraction of biomass) producing only electricity. These traditional HES can be characterized as MISO (multiple inputs, single output) systems. On the other hand, HESi denote energy configurations producing electricity as well as other energy products Pi. These HES configurations can be characterized as MIMO (multiple inputs, multiple output) systems, with these multiple outputs enabling an immediate alternative use of excess energy. This feature allows HES to act as dispatchable generation centers (a characteristic highly valued by utilities) that can be reliably called into service when required, even under the presence of undispatchable generation from renewable sources. Furthermore, by absorbing the renewable variability itself via locally-coordinated and controlled energy conversion sub-systems, HES can potentially make the grid fundamentally immune to ill-effect of renewable variability to a large extent. The additional opportunities for energy management offered by MIMO HES can be exploited to reduce variability costs, increase profitability, and improve operational flexibility and electrical response.

Fig. 1. Energy grid topology with traditional generations (Gi), hybrid energy systems (HESi) and loads (Li). In addition to electricity, HES may also produce other products (e.g., synfuel (P1), methanol (P2), ammonia (Pm)).

4

H.E. Garcia et al. / Energy 52 (2013) 1e16

Fig. 2 illustrates how HES can provide opportunities for accommodating the variability introduced by variable renewable generation. In particular, assume that the power grid operator (e.g., an ISO (independent system operator), where a given HES is located) requires that this HES delivers the generation profile GP4, a flat (electrical) generation profile. Assume also that the REN (renewable sources) generate GP1, a highly variable generation. Given that GP1 is different from GP4, an ESE (energy storage element) may be employed to smooth GP1, making it closer to GP4. Because the smoothed GP2 is still different from GP4, the PHG (primary heat generator) must generate GP3 so that when added to GP2, their combined contribution matches GP4 (i.e., GP2 plus GP3 equals GP4). Examining GP3, it can be noticed that PHG must thus be run under flexible operation, as opposed to base-load operation. If chemical products can also be produced at a CPC (chemical plant complex) under an advanced MIMO HES, a GP5 becomes available in order to best distribute and manage the variability introduced by renewable sources among HES constituents. Because now GP4 equals GP2 plus GP3 minus GP5, an additional opportunity exists under the advanced HES to deal with GP1’s (and possibly GP4’s) variabilities. Recognizing the tight interdependency among supply, transmission, distribution, and demand, an implicit feature of the energy grid topology illustrated in Fig. 1 is that each HESi participates in the existing electricity power generation market attempting to deliver the electricity profile assigned to it by a regional power grid operator, while, at the same time, independently producing other energy products Pi. As typically done, this assigned electricity profile is based on a predicted demand curve and market bids received from interested utilities [40,41]. Thus, from an electricity market perspective, the power grid operator is essentially only interested in that a given HESi closely deliver the assigned electricity generation profile. What a given HESi may do with any potential excess of energy production lies within the (internal) interests of the particular HESi. HES so conceptualized can be integrated smoothly with existing infrastructure and be a key strategy to electrical grid modernization. Their particular energy arrangement can also significantly improve knowledge of energy flows within themselves and the power grid. 1.3. Proposed methodology Issues concerning flexible operation and variable renewable generation have been addressed in the literature. For example in Ref. [42], six main options are discussed for managing the

Fig. 2. Distribution of renewable generation variability among HES components, with GPi denoting the “i” generation profile.

renewable variability, namely, power plants providing operational and capacity reserves, electrical storage, interconnection with other grid systems, distributed generation, demand-side response, and curtailment of intermittent technology. Among these options, the most popular is using flexible power plants with capacity reserves [42]. The capacity reserves of power generating units collectively provide a “spinning reserve” for power grid operators to use in case of a generationedemand imbalance. For example, natural gas-fired turbines have been used to smoothen renewable variability due to their quick start up time and relatively faster ramp rate [30,43]. In some of the modern power generation system, this variability is also mitigated by distributed generation and hydropower resource [44e46]. Energy storage is another way to mitigate variability in renewable generation, but most commercial ESE technologies are expensive and inefficient with low life cycle [47e49]. In general, our literature review suggests that most research efforts emphasize time series or statistical analysis when analyzing the effect of renewable variability on the power grid and suggesting variability mitigation strategies. Little emphasis has been given in examining the system behavior in a dynamic framework of closely-coupled energy producing and consuming units under high renewable penetration. This is the broad objective of this work: analyzing this dynamic performance and the associated cost effects. A number of ways to model and optimize the dynamic behavior of hybrid systems, particularly those that operate in support of a power grid rather than as a complete microgrid, have been reported. Classically, a year’s worth of wind generation data (either as wind speed fed to a turbine generator model, or as generator output directly) at hourly intervals is supplied to a system model which integrates the performance over a year to either predict a particular system’s expected average performance or to select equipment sizes or types to obtain optimal performance [50]. The value of direct dynamic simulation, as opposed to the time-averaging of steady-state performance at various states, is noted in Ref. [51]. They acknowledge the greater complexity and computation time of dynamic simulation and go on to discuss a way of generating non-uniform Time slices to best perform the step-wise determination of average performance. The complexity of performing optimizations of a multivariable system is discussed in Ref. [52] and further discussed for multi-objective optimizations of heat and power generation in Ref. [53]. Subtler costs of operation resulting from excessively rapid change of state were considered [52]. Fuzzy logic has also been applied to the optimization problem [54]. The analysis presented here differs from these efforts in the sense that the comparative findings reported here are derived from using results computed from dynamically simulating and operating HES consisting of multiple, tightly-coupled dynamic constituents. Consequently, dynamic models of each HES constituent are developed and accordingly integrated within different HES configurations. While the level of modeling detail is often basic (i.e., first-order linear differential equations), greater modeling details (including nonlinearities) are included when deemed necessary for sufficiently characterizing certain dynamic components. Among the parameters included in these dynamic models, a subset of them istreated as variables in order to determine not only the consequences of varying them but also their best values for achieving given dynamical requirements under indicated constraints. The identified best values for these tunable parameters provide functional requirement guidelines for designing HES components and equipment, without necessarily suggesting how to achieve these best values or desirable dynamical characteristics. A dynamic simulation capability was developed for integrating component models, solving their dynamic characterizations, computing synthetic results, and analyzing findings. It is argued that the operational implications resulting from the variability introduced by not only time-varying electricity demands

H.E. Garcia et al. / Energy 52 (2013) 1e16

but also renewable integration can only be effectively understood in a dynamic setting. Such a dynamic analysis is carried out by introducing HES that may accept several energy sources (e.g., nuclear, carbon-based sources, and renewable) and produce one or more energy commodities, such as electricity and chemical products (e.g., synfuel and ammonia). This enhanced understanding will assist policy makers as well as engineers to best devise practical solutions for accommodating the high variability arising from higher renewable energy penetrations.

5

describes different operational scenarios. In Section 3, definitions of technical terms used in analyzing HES are provided and explained. In Section 4, models and time-series data representing various components of HES are described. In Section 5, dynamic performance analysis results are presented for candidate traditional and advanced HES and their implications are discussed. Section 6 concludes this paper.

2. HES configurations 1.4. Technical contributions Preliminary dynamic performance analyses are conducted to evaluate HES solutions under flexible operation and variable renewable generation. HES solutions are demanded to deliver a given electrical generation profile. This profile is assigned by the power grid operator (e.g., an ISO) based on a predicted demand curve and bids received from power grid utilities. The given electricity profile must be delivered regardless of increasingly levels of renewable penetration. The main contributions of this initial dynamic performance analysis are as follows:  compare, in a dynamic setting, traditional versus advanced HES candidates;  conduct dynamic performance analysis of HES in order to investigate opportunities and challenges related to increasingly higher penetration of renewable energy and effects that dynamic characteristics and potential synergies among HES constituents may have on the overall dynamic performance of HES candidates, in mitigating variability, and in stabilizing power grid operations;

1.5. Manuscript organization The rest of this paper is organized as follows. Section 2 discusses the topological architecture of considered HES configurations and

HES can have diverse network topologies and purposes. For this work, candidate HES are envisioned to be microgrids connected to the power grid. The energy configuration assumed here for an advanced HES is illustrated in Fig. 3, which includes the following nine main components:  a 150-MWe heat generation plant (e.g., a small modular reactor) that generates steam utilized for different purposes (e.g., to generate electricity and chemical products), denoted as the PHG (primary heat generator);  a steam turbine and electrical generator pair that converts steam into electricity, denoted by the STG (steam turbine generator);  a fast, load following power plant (e.g., a prime mover) that provides auxiliary electricity generation, denoted by AEG (auxiliary electrical generation). Although modeled, this unit is not utilized in this initial analysis;  a renewable power generation capability of up to 60-MWe (e.g., 20  3-MWe capacity wind turbines), denoted by REN (renewable);  an electrical storage that smoothens the electricity generated by the renewable source, denoted by the ESE (energy storage element);  an auxiliary heat generation plant of up to 60-MWe (e.g., a natural gasefired boiler) that generates additional ondemand steam, denoted by AHG (auxiliary heat generation);

Fig. 3. Architectural topology of the considered advanced HES.

6

H.E. Garcia et al. / Energy 52 (2013) 1e16

 a heat transfer interface that dampers fluctuations in the steam produced by the heat sources before it is used for chemical production, denoted by HTE (heat transfer element);  a chemical (e.g., methanol) production plant able to utilize heat equivalent to up to 60-MWe, denoted by CP (chemical plant);  enough carbon (and hydrogen) sources (e.g., NG) to support chemical production. The primary source of heat in these HES is PHG, which can be either a nuclear fuel- or fossil fuel-based steam production plant. There are two additional units for electricity generation, namely, AEG (auxiliary electrical generation) and REN (renewable), with the latter coupled with an ESE. These three electricity generating units are operated accordingly to deliver the required electricity generation requested by the power grid operator (e.g., an ISO, RTO). This collective electricity production is the first output of the HES system. The second output is chemical products (e.g., methanol), which is produced from utilizing carbon sources (e.g., NG (natural gas)) and steam generated by PHG and, if needed, AHG. The CPC (chemical plant complex) consists of CP, HTE, AHG, and a carbon source to drive AHG and chemical processing plant. As suggested in Ref. [55], the strategy for efficient energy utilization is to build the PHG with a capacity that meets the peak electrical load, with excess process steam diverted to the production of chemicals or other products. By producing more than one energy product, the candidate HES adds additional opportunities for flexible energy management, essentially acting like a large energy smoothing vehicle for the grid. A large PHG with sufficient capacity also provides robustness to the overall stability of the grid as electrical demand can be met by this HES irrespective of whether renewable energy is present. This robustness is vital for the synergetic integration of dispatchable, controllable, predictable, and low-variability energy units with undispatchable, uncontrollable, unpredictable, and high-variability renewable sources. While the nondispatchability characteristic of renewable generation is a difficult technical challenge that may be partially tackled via better forecast models, the high variability of renewable sources may be somewhat mitigated using ESE. The CP utilizes excess steam from the PHG to produce energy-storing chemical products. By generating additional steam to compensate for variability in the PHG excess steam, an AHG enables the chemical plant to operate at a given constant (or low variability) production mode. Fig. 3 indicates that energy flows among HES components in two types, namely, either electricity or heat in the form of steam. Based on thermodynamical and efficiency calculations embedded within our STG (steam turbine generator) model, it was estimated that 1 [kg sec1] of steam produces 0.771 MW of electricity (i.e., 0.771 MWe). The assumed steam inlet condition for this calculation is 6 [MPa] and 275 [ C] (saturated). Given this relationship and for simplicity, all power calculations will be expressed using the electrical equivalence (in MWe) of the particular power stream. For example, consider that the HES need to deliver a flat electrical generation profile of 150 MW and that the PHG operates at 150 MW constant baseload unit. If the renewable sources generates 60 MWe at a given time, this additional generation will translate into additional power going to the CPC. This additional power contribution is expressed as 60 MW of equivalent electrical power (i.e., 60 MWe), as opposed to 78 [kg∙sec1] of additional steam. It is important to point out that Fig. 3 makes the simplification of assuming that heat in the form of steam from the PHG could be used not only for electricity generation but also for steam methane reforming of natural gas to produce synthesis gas for methanol production. Considering that it typically operates at a temperature of around 650  C or higher, this assumption implies that the steam reforming reactor would need to be redesigned for convective

(instead of radiation) heat transfer, for example, which may increase its cost and require significant development from current industrial technologies. In addition, while power cycles for electricity production can be designed to use steam at a temperature of around 650  C (though typically being closer to 600  C), concerns may arise from using steam above 550  C at the PHG, particularly if it is a nuclear reactor. Hence, while producing higher temperature steam in order to accommodate the needs of the considered chemical process could lead into significant technological challenges, this reference energy architecture was selected in order to keep the dynamic analysis presented in this paper at an adequate level of simplicity.

3. Definitions 3.1. Energy accommodation EA (Energy accommodation) is the ratio of steam coming from PHG that is actually introduced (not vented) and utilized by the CPC to directly produce chemical products. Steam rejection is introduced to reduce the variability of the steam flow rate coming from PHG. EA is determined based on the maximum flow rate (Fcp_max) and the steam profile utilized by CPC. Given a magnitude for EA, a search procedure is utilized to determine Fcp_max that produces the given EA. This value for Fcp_max is then used to reject any steam rate that is higher. While this operating strategy reduces steam flow variability, the lower EA is, the more rejected (and unused) steam there is, hence less steam for directly producing chemical products.

3.2. Required generation RG (Required generation) is the electricity generation profile that must be delivered to the grid by the HES. This profile is typically assigned by an ISO or a RTO (Regional Transmission Organization). In general, ISO/RTO may solicit a time-varying generation profile; however, its variability is significantly lower and more predictable than that expected for renewable energy sources. The inherent assumption though is that the requested generation profile can be met by the participating energy systems without exceeding their allowable ramping up/down rates. For simplicity, a flat, constant electricity generation profile of 150 MW is assumed hereafter unless otherwise indicated. However, the dynamic analysis results discussed here can be readily extended to more elaborate generation profiles.

3.3. Renewable penetration RP (Renewable penetration) is defined using the following relationship:

h i max E_ ren ðtÞ % Renewable penetration ¼

t0

max ½RGðtÞ

 100;

(1)

t0

where E_ ren ðtÞ is the renewable power generation (see Fig. 9) and RG(t) is the required generation at time t. This particular formulation for RP was selected over other possibilities to avoid assigning an ill-defined value whenever the denominator RG(t) is zero at a given instant t. It is important to indicate that Eq. (1) applies to the microgrid defined by the given HES, not to the regional grid as a whole.

Renewable power generation (MW)

H.E. Garcia et al. / Energy 52 (2013) 1e16

7

30 25 20 15 10 5 0 90

91

92

93

94

95

96

97

Days Fig. 4. Power generation profile used to model renewable power (i.e., one week of synthetic wind generation data sampled every 10 min from early April for a site in Wyoming).

3.4. Achievable generation and average generation deficit Given RG (as requested by the power grid operator for a particular HES), the AG (achievable generation) is defined as the generation that can be realized by the HES system without violating the operational constraints delimited by the maneuverability values of the constituent energy units. The AGD (average generation deficit) is computed using the following relationship:

1 AGD ¼ DTop

D ZTop

½RGðtÞ  AGðtÞdt;

(2)

0

where DTop is the time window considered of HES operation. Notice that in the formulation above for AGD, the absolute value operator is not used in the integrand as AG(t) is computed in a manner that assures AG(t)  RG(t) at all times. 4. Model description A unique characteristic of renewable energy is its high variability. Components closely interacting with variable renewable generation need to be able to accordingly maneuver their operations in order to best accommodate this variability. This section introduces the need for all energy components of the considered HES and their dynamic models characterizing the desired timescale dynamic behavior with appropriate fidelity. This section also introduces several notions of maneuverability depending on which energy component is considered. Maneuverability here refers to the agility of a system to respond to its inputs. Agility constraints may result from inherent physical limitations, from safety mechanisms implemented in order to meet safety requirements, or from operational procedures intended to reduce performance (and economical) degradation during transients, for example.

4.1. Renewable energy The renewable energy is treated as a must-take input to the HES system under consideration. As shown in Fig. 3, it might be introduced to HES via an ESE (electric storage element) or directly. It denotes energy provided by different renewable energy sources, such as wind and solar, and usually characterized by high variability, intermittency, and nondispatchability. The non-scheduled, weather-driven renewable energy sources show variability in all time scales. For long term prediction of wind power, physical models may be utilized; however, they often provide insufficient accuracy in short-term for efficient power management. Likewise,

while statistical methods may provide accurate short-term results, their reliability in long-term is questionable [56]. In this work, renewable energy generation is modeled using synthetic time-series data of wind power generation obtained from the Western Wind dataset created for the NREL (National Renewable Energy Laboratory).1 Data for about 32,000 possible wind generation sites around the western United States are available. That dataset assumes ten 3 MW capacity wind turbines at each 2 square kilometer site for a maximum of 30 MW generation at full production. Considered typical among regions with high wind resources, the data series used in this work is for a site in Wyoming with a high capacity factor of 40%, meaning that, because of the variability of wind speed, over a year the turbines produce 40% as much energy (in MWh) as if they had run continuously at full output. However, the dynamic analysis here presented can be readily applied to more typical regions with low capacity factors. Fig. 4 shows a representative seven days of the dataset (sampled at every 10 min, with linear interpolation used to fill data gaps) to illustrate the variability of power production which must be accommodated to make wind power more generally useful. Scaled versions of this time-series are utilized to model different levels of renewable penetrations. It must be noted from Eq. (1) that renewable penetration is computed using the maximum value observed in the renewable energy generation profile considered as opposed to its average. For wind, the average renewable energy delivered is about 30e40% of its maximum instantaneous value. For example, the renewable power profile shown in Fig. 4, with 30 MW peak power, corresponds to an average power generation of 12.4 MWe. This implies that if the maximum RG for a given HES is 150 MWe, then a 40% renewable penetration corresponds to a 60 MWe maximum renewable power (which would demand a 60 MWe of installed wind capacity) and to a 18e24 MWe average generation. Other renewable sources, such as solar, geothermal, and hydro, are not considered in this study; nevertheless, these energy sources may be treated as AEG units or as wind units with a different time series of generation within the developed simulation framework for analysis purposes.

4.2. Heat generation Heat generation units are introduced in the considered HES as primary source of energy to produce process steam used for two purposes: (1) To drive a steam turbine and generator pair in order to produce the required additional electricity for meeting grid

1

Accessed on 20 April 2010 at http://www.wind.nrel.gov/Web_nrel/.

8

H.E. Garcia et al. / Energy 52 (2013) 1e16

demand considering contribution from renewable; (2) To provide process heat to chemical plant for synthesizing chemical products from natural gas. As shown in Fig. 3, the considered HES include two heat generation units, namely, PHG (primary heat generation) and AHG (auxiliary heat generation). The PHG is sized for full loadoperation, therefore capable of generating by itself (hence without renewable contribution) sufficient process steam to meet the anticipated maximum electrical grid demand. Under such an extreme situation, no process steam generated by PHG is directed to the chemical plant, whose needs would have to be solely fulfilled by the AHG unit; in the event of non-zero renewable contribution, the excess steam produced by the PHG operating at full-load is sent to chemical plant. If this excess steam coming from PHG is not sufficient to maintain the desired production level at the chemical plant, AHG is accordingly regulated to provide the required additional steam. In summary, the PHG provides steam for electricity generation as well as chemical plant use, whereas AHG provides steam exclusively for chemical production. It is assumed that these heat generation units provide process steam at a constant temperature and pressure (6 [MPa] and 275 [ C], saturated). The mathematical model of heat generation illustrated in Fig. 5 is an adaptation of a moderately complex nonlinear superheater boiler model proposed in Refs. [57,58]. For the analysis presented here, this model captures essential dynamic characteristics of a drum-boiler over the desired range of operations. The boiler pressure is controlled using a feedback linearization-based nonlinear controller and the steam flow is controlled using a PI (proportionaleintegral) controller on the steam valve actuator. This heat generation model is based on simplifying assumptions so that the long term simulations of the integrated model can be completed in a reasonable time frame. The pressureetemperature effect on the steam production rate KT ¼ (vTsat/vpsat) and pressureeenthalpy effect on steam rate KH ¼ (v(hg  hfw)/vp) are not significant and therefore, were ignored in this work [57]. The parameters KSH, TF, CB, SF, SG, and s represent superheater friction drop coefficient, fuel and waterwall time constant, boiler storage constant, turbine steam flow, steam generation, and fuel and waterwall time delay, respectively. The controller parameters Kp and KI represent the proportional and integral gain of a PI controller. In term of complexity, the model is simple, yet adequate enough to represent important dynamics associated with actual plants. With respect to PHG and AHG, maneuverability essentially refers to its load following capability, which may be imposed by diverse factors, including inherent physical time constants, non-

Fig. 5. Nonlinear model of a drum steam boiler.

instantaneous response of operators, controls, and safety requirements. Advanced designs for flexible operation, modern instrumentations, and skilled human operators may result in faster responding units; however, from a cost point of view, it is usually more economical to operate these units at a flat generation profile. In order to simulate different load following characteristics, the valve-opening actuator regulating fuel utilization is modeled as a rate-limited unity gain. An associated controller is designed so that the rate of control action meets the limit imposed by the ratelimited actuator. By changing this rate-limit, referred here as ramp rate limit, different load following characteristics can be simulated for PHG and AHG. The ramp rate limit for PHG, i.e., Mphg , and AHG, i.e., Mahg , is modeled by the corresponding rate-limit rmax of the valve actuator shown in Fig. 5. 4.3. Steam turbine generator The STG (steam turbine generator) unit is the primary source of electricity generation in the considered HES. As shown in Fig. 3, the HES is operated in a manner such that the given grid demand is met from accordingly combining the electricity produced by STG, renewable generation via the electric ESE (energy storage element), and AEG (auxiliary electrical generation). A nonlinear electrical production model is developed for steam to electrical conversion consisting of following four components: steam throttle valve, a three-stage tandem compound system including a single reheat turbine with steam extraction from highpressure turbine, generator, and load component. The model assumes supply of steam at a constant pressure. Unlike other steam turbine-generator models for power system modeling [59], complete nonlinear modeling of steam quality, and enthalpy are here computed from thermodynamic data along with first-order transfer function modeling of steam chest, inlet piping, reheater, and crossover piping. Classical electric/hydraulic governor and AGC (automatic generation control) system are implemented to control the steam throttle valve. Fig. 6 shows the considered steam turbine configuration. Fig. 7 shows the modeling diagram for the schematic of the turbine system shown in Fig. 6. All three stages of this turbine are modeled as pressure expander with fixed pressure drop determined by the design. The justification regarding this assumption can be made as follows. In a complex energy system, such the one addressed in the current research, the dynamic characteristics can be considered at various levels. Two most important levels are: dynamic power generations and dynamic thermo-hydraulics. In this paper, we have considered dynamic power generation. Even though dynamic thermo-hydraulic simulation is more detailed than dynamic power generation, dynamic power generation is one of the industry-standard dynamic simulation approaches for

Fig. 6. Schematic of a compound tandem, single reheat turbine system configuration. _ and x represent pressure, temperature, rate, and quality of steam flow, Variables p, T, m respectively.

H.E. Garcia et al. / Energy 52 (2013) 1e16

9

Fig. 7. Block diagram of developed reheat turbine system model.

energy systems [59]. As in Ref. [59], the dynamic nature of the model is mainly due to varying flow rate through various system components. The operating points of components, such as, inlet pressure and outlet pressure of turbines, are either assumed to be constant or controlled at a fixed set-point. The work performed by each stage of this turbine can be determined from thermodynamic considerations. Let the input steam condition of the ith stage, i ¼ 1, i ; xi Þ, where xi is the steam quality 2, 4 (see Figs. 6 and 7) be ðpiin ; Tin in in at ith turbine stage inlet and the output steam pressure be piout ¼ piin  Dpi , where the constant Dpi > 0 is the designed pressure drop across the turbine stage. Assuming isentropic steam expansion, the isentropic enthalpy at ith stage outlet can be found out to be Ziout ¼ hðsiin ; piout Þ, where siin ðpiin ; xiin Þ is the inlet entropy of the steam. The isentropic efficiency of turbine stage is defined as follows:

hiise ¼

hiin  hiout hiin  Ziout

;

(3)

where hiout is the actual outlet enthalpy. The isentropic turbine efficiency (hiise < 100%) is assumed to representative of mechanical efficiency of turbine, heat loss in turbine, generation efficiency of generators, and other losses in a lumped manner. From Eq. (3), the enthalpy of the steam at the outlet can be written as follow:

hiout ¼ hiin  hiise



 hiin  Ziout :

Pmech ¼

i ¼ 1;2;4

_ iin m

h

hiin



hiout

i

4.4. Heat transfer element A HTE (heat transfer element) facilitates transfer of heat energy in the form of process steam to the chemical plant. An HTE may correspond to an intermediate heat exchanger, for example. As shown in Fig. 3, this energy component is placed between process steam coming from PHG (primary heat generation) and AHG (auxiliary heat generation) and the chemical plant. This energy component is necessary to bring the conditions (pressure and temperature) of the incoming steams from PHG and AHG to acceptable values as required by the chemical plant. Apart from this basic task, there are other advantages of having a heat exchanger in the loop, especially when the PHG is a nuclear power plant, such as creating a safe barrier immune to off-normal events at chemical plant end, providing control over steam operating temperature at chemical plant end, mitigating tritium transport [60e62], and, most important for this study, attenuating the potential variability in the process steam received at the CPC from the PHG. The HTE is modeled as a first order linear system represented by the following transfer function [63]:

(4)

The steady-state mechanical power imparted by the turbine can be expressed as the enthalpy difference as follows:

X

model discussed above whenever deemed appropriate for meeting computation time constraints. It must be noted that this constant value depends on inlet steam condition, operating condition, and characteristics of the modeled STG unit.

¼

_ iin hiise m



hiin



Ziout



;

(5)

HTEðsÞ ¼

1

shte s þ 1

;

(6)

where shte is the time constant characterizing the damping action conducted by HTE. 4.5. Chemical plant

_ in is the mass flow rate of steam through the ith turbine where m stage. Appropriate transfer functions are used to represent dynamics associated with subcomponents, such as crossover, reheater, and steam chest (low and high pressure). It is important to indicate that after multiple simulations and detailed analysis, it was confirmed that the developed STG dynamic model exhibits the fastest dynamics among all considered HES constituents. For example, the response time of a turbine-generator units can be 10e 100 times less than other units (e.g., chemical plant). Consequently, a constant gain was utilized in lieu of the complex full dynamic

As shown in Fig. 3, CP (chemical plant) utilizes process steam produced by the PHG (and possibly complemented with AHG steam) to convert carbonaceous fuels such as NG (natural gas) into chemical products (e.g., methanol, diesel). It is assumed that a steam reforming reaction is taking place in the chemical plant producing synthetic fuel (e.g., methanol). Fig. 8 describes the model _ scp ðtÞ is the input steam mass flow used for simulating CP, where m _ ccp ðtÞ is the chemical plant required carbonaceous mass flow rate, m rate, scp > 0 is a parameter characterizing the maneuverability of

10

H.E. Garcia et al. / Energy 52 (2013) 1e16

removing the variability introduced due to renewable penetration. The effect of this smoothing action is modeled as illustrated in Fig. 9, where E_ ren ðtÞ is the renewable electricity generation profile f (e.g., produced by wind generators) and E_ ren ðtÞ is the filtered renewable electricity generation profile directly contributed by renewable sources to meet the required electricity generation profile assigned to the given HES by the power grid. A filtered f electricity profile E_ ren ðtÞ can be delivered to the grid by timely taking power (e.g., charging a battery) and delivering power (e.g., discharging a battery) from/to the incoming electricity generation E_ ren ðtÞ. A first order linear system is used for modeling ESE as follows [64]:

Fig. 8. Chemical plant model.

the chemical plant, Kcp represents the static gain of the chemical _ pcp ðtÞ plant from a steam-input-product-output point of view, and m is the product production rate. As shown in this figure, it is assumed that, in addition to 1 kg of steam, 0.175 kg of NG is needed to produce 0.35 kg of methanol. These relationships are estimated after solving thermalechemical formulations and assuming a chemical production yield (CP yield) hcp of 100%. The transfer function representation of the chemical plant can be written as follows:

CPðsÞ ¼

Kcp

scp s þ 1

:

(7)

While the yield for steam reforming and methanol synthesis is high, it may be deemed optimistic to assume in practice a carbon efficiency of 100% for converting methane to methanol. If CO2 is added, more than 1 mol of methanol can be produced per mole of CH4. However, this requires a high purity source of CO2, which is not typically done commercially. Considering that yields for chemical plants may be lower than 100%, results are also here computed assuming 90% and 80% yields. Furthermore, it is assumed that the _ ccp ðtÞ is controlled so that mass flow rate of carbonaceous material m s _ ccp ðtÞ is always _ cp ðtÞ and m the stochiometric ratio between m maintained to produce the energy-storing products in the chemical plant. The CP time constant scp dictates how slow or fast the CP responds. The response may be affected by numerous factors including thermal or mass accumulation and mixing phenomena or the rate of building up of key reactants. However, as the CPC consists of three dynamical components, namely, AHG, HTE, and the CP, the maneuverability Mcpc of the CPC is collectively dictated by not only scp but also the ramp rate limit of AHG (Mahg ) and the Þ. Recall that an AHG is maneuverability of HTE represented by ðs1 hte utilized to produce additional steam so that when added to the stream flow coming from PHG the resulting steam flow going to the CP exhibits less variability than that of the incoming PHG steam. Thus, in the dynamic analysis here discussed, a given value for the chemical plant complex maneuverability will not correspond to the maneuverability for the chemical (methanol production) plant but rather to the maneuverability for the chemical plant complex as a whole. This remark is included as chemical plants are usually run under very tightly controlled conditions. This operational strategy is typical because if operating conditions vary substantially, product yield, catalyst life, and plant output rate can be significantly reduced, in addition to possibly compromising plant safety and lifetime.

ESEðsÞ ¼

1

sese s þ 1

;

(8)

where s1 ese is the smoothing frequency associated with ESE. Notice that as modeled, the smoothing frequency of the ESE fese ¼ s1 ese determines the renewable variability attenuation effect achieved from using an ESE, thus defining the degree of variability removal (or degree of smoothing) that the electrical renewable contribution to the power grid would experience from its original profile coming directly from wind generators, for example. The smaller the smoothing frequency, the higher the variability attenuation. A higher variability attenuation is in turn realizable using an ESE with a higher capacity. For example, increasing s1 ese would decrease the smoothing (filtering) effect that the ESE would have on the incoming renewable generation profile, consequently also decreasing the difference between the input renewable generation and its output contribution to the power grid. As this difference is the electrical signal that must be accommodated by the ESE, less inputeoutput difference between these renewable generation profiles implies less maximum storage capacity (in MWh) that would be required for the ESE as indicated in Fig. 9. The smaller the smoothing frequency, the larger maximum energy capacity required for ESE. The corresponding required maximum capacity is computed from identifying the half cycle with maximum area within the inputeoutput difference signal. On the other hand for simplicity, no limitation is imposed on the maximum rate (in MW) of uptake (e.g., charge) or delivery (e.g., discharge) of the ESE.The modeling of ESE as a low pass filter can be explained as follows. Assume that the ESE is a pumped hydro storage element in which the energy is stored as potential energy of water pumped to a storage lake at high elevation. The renewable energy source is not connected directly to the grid but instead runs a pumping station that keeps the storage lake full. This water can be released to

4.6. Energy storage element The ESE (energy storage element) may be the first mechanism for mitigating the high variability introduced by renewable energy sources. Some examples of such ESE are batteries, CAES (compressed air energy storage), and ultra-capacitors. The interest lies here with the smoothing action that the ESE may perform in

f

Fig. 9. Storage model. The plot of ðE_ ren  E_ ren Þ is not drawn to scale.

H.E. Garcia et al. / Energy 52 (2013) 1e16

renewable variability on its own. However, this solution may be impractical. For example, consider a wind turbine station with supervisory ESE controller set-point fixed at 20 MW. Furthermore, consider a day when there is only enough wind energy to produce 8 MW of power. Then, on an average, the ESE would have to be sized for capacity ¼ (20  8)  24 ¼ 288 [MWh]. The largest battery storage in world is only 36 [MWh] [65]. In summary, the smoother the desired output of the ESE, the larger its storage capacity must be, which may be technologically or economically prohibiting.

70

Average Generation Deficit (MW)

Ramp rate limit [hr−1] 60

50

± 1/48 ± 1/24

Better load following

40

11

± 1/12 30

5. Dynamic performance analysis

± 1/3 20 ± 1/1 10

0

± unlimited

0

5

10

15

20

25

30

35

40

Renewable penetration (%) Fig. 10. Effect of renewable penetration on average generation deficit at various PHG ramp rate limits.

a lower elevation lake through turbines driving generators that are dispatched at rate determined by a supervisory controller or by the grid director, for example. The controller set-point is changed depending upon the current level of wind generation and average short-term wind prediction. This approach can help maintain a moderate rate of ESE output power variability even though the renewable energy might have high variability. So an ESE, along with a well-performing supervisory control, may be able to remove high variability generation from renewable sources. This performance is similar to low-pass filtering action. Similar analogies can be made for other ESE technologies (e.g., a battery). This desired performance will be limited by the size of the storage system. If the set-point for the ESE controller were fixed at a constant value and a constant output were dispatched to the grid, it might then appear that ESE can completely solve the problem of

In this section, dynamic analysis results are discussed, including investigating the effects that various dynamic factors and properties have on achievable generation. In particular, the effects of varying the PHG (primary heat generation) maneuverability or load following characteristics of PHG (i.e., Mphg ), the storage smoothing frequency (i.e., fese), and the CPC (chemical plant complex) maneuverability (i.e., Mcpc ) are discussed. Findings corroborate the conjecture that the overall dynamic characteristic of a given HES configuration depends on the individual dynamic properties of each constituent, the dynamic interaction/coupling among them, and the synergies that can be exploited. Thus, this dynamic analysis provides insights to devising optimized HES for integrating different energy units as well as strategies for meeting required generation profiles while enabling increasing penetration of renewable energy. In order to generate the time series data needed to conduct this dynamic analysis, a simulation capability was developed that solves the dynamic models using the DormandePrice method with a variable time step and a relative tolerance of 107. 5.1. Effect of PHG maneuverability (ramp rate limit) on achievable generation In principle, both PHG and CPC can be accordingly operated to accommodate variability in an HES configuration. However, even though it may not be advisable based on cost, profitability, and safety considerations for PHG to undergo highly variable operation,

Renewable Penetration =20% 160 150

Achievable Generation (MW)

140 130 120 110 100

Required Generation

Mphg= ± unlimited

Mphg= ± 1 hr−1

−1

Mphg= ± 1/48 hr

90

Renewable Penetration =40% 160 150 140 130 120 110 100 90 50

51

52

53

54

55

56

Days Fig. 11. Effect of renewable penetration on achievable generation at various PHG ramp rate limits as function of time.

57

12

H.E. Garcia et al. / Energy 52 (2013) 1e16

70

Average Generation Deficit (MW)

Smoothing frequency [hr−1] 60 No storage 50

2 More storage 1

40

0.5

30

0.25 20 0.10 10

0

0

5

10

15

20

25

30

35

40

Renewable penetration (%) Fig. 12. Effect of renewable penetration on average generation deficit at various storage smoothing frequencies.

the dynamic response behavior can be specified for both these units without requiring that they be operated dynamically. We believe that steady operation is the preferred mode, but this model includes the ability to vary these units’ rates to see if modest rate swings (and perhaps infrequent large rate) offer any advantage. On a side note, some of the new nuclear power plants in Europe (hence nuclear based-PHG) are currently operated in limited loadfollowing mode. In the near future, it may be possible to operate reactors in a more aggressive load-following mode [66,67] with the advent of SMR (small modular reactors), for example. Keeping this fact in perspective, AGD (average generation deficit) is evaluated under several values of PHG ramp rate limit and renewable penetration. The values of other parameters used in this scenario simulation are as follows: chemical plant complex maneuverability Mcpc ¼ 0:36 [hr1] and fese ¼ N [hr1] (i.e., no storage). In particular, Fig. 10 shows the average generation deficit

as a function of renewable penetration level for various PHG ramp rate limits, the latter essentially corresponding to the load following characteristics of the given PHG. These results suggest that: (1) for a given renewable penetration level, as the PHG ramp rate limit increases (i.e., better load following), the average generation deficit decreases; (2) for a given PHG ramp rate limit, as the renewable penetration increases, the average generation deficit increases, hence progressively deviating from the required generation profile. In other words, PHG with better load following capabilities facilitate achieving the required generation profile under variable renewable generation. It is important to notice from Fig. 10 a monotonically increasing AGD (average generation deficit) as RP (renewable penetration) increases, even under an unlimited PHG maneuverability (or ramp rate). This deficit is due to the limiting dynamics characteristic imposed by CPC due to its finite maneuverability. Fig. 11 shows a seven day time snapshot of achievable generation assuming three different values for PHG ramp rate limit (i.e., Mphg ¼ N [hr1], 1 [hr1], (1/48) [hr1]) and at two different renewable penetration values (i.e., 20% and 40%). These results suggest that using increasingly faster responding PHG may allow the accommodation of more renewable energy. As mentioned previously, the chemical plant complex maneuverability does not necessarily correspond to the maneuverability of the chemical plant by itself but rather of the combined CP (chemical plant) and AHG (auxiliary heat generation). Therefore, when Mcpc ¼ 1=36 [hr1] is assumed for this scenario, it does not necessarily imply that CP is operated at a maneuverability rate of 1/ 36 [hr1]. In fact for this case, it was assumed that the CP is operated under steady state mode (Mcp ¼ 0 [hr1]), while AHG is operated at (Mahg ¼ 1=36 [hr1]), hence the combined maneuverability of 1/36 [hr1]. 5.2. Effect of storage on achievable generation AGD is evaluated under several values of storage smoothing frequency and renewable penetration, assuming the following parameter values: PHG ramp rate limit Mphg ¼ ð1=48Þ [hr1] and CPC maneuverability Mcpc ¼ 0:36 [hr1]. In particular, Fig. 12 Renewable Penetration =20%

150

Achievable Generation (MW)

100

50

Required Generation

fese= 0.10 hr−1

−1

fese= 0.25 hr

0

No Storage Renewable Penetration =40%

150

100

50

0 50

51

52

53

54

55

56

Days Fig. 13. Effect of renewable penetration on achievable generation at various storage smoothing frequencies as function of time.

57

H.E. Garcia et al. / Energy 52 (2013) 1e16

5.3. Effect of chemical plant complex maneuverability on achievable generation

60 −1

0.36

Average Generation Deficit (MW)

CPC Maneuverability [hr ] 50

1.8

40

Faster chemical plant complex (CPC)

30

3.6

20

7.2

10

14.4

unlimited

0

−10

0

5

10

15

20

25

30

35

13

40

Renewable penetration (%) Fig. 14. Effect of renewable penetration on average generation deficit at various chemical plant complex maneuverability values.

shows the average generation deficit as a function of renewable penetration level for various storage smoothing frequencies. These results suggest that: (1) for a given renewable penetration level, as the storage smoothing frequency decreases (more storage), the average generation deficit decreases; (2) for a given storage smoothing frequency, as the renewable penetration increases, the average generation deficit increases, hence progressively deviating from the required generation profile. In other words, more storage facilitates achieving the required generation profile under variable renewable generation. Fig. 13 shows a seven day time snapshot of achievable generation assuming three different values for storage smoothing frequency at two different renewable penetration values. These results indicate that a potential solution to accommodate more renewable energy is to increase storage.

AGD is evaluated under several values of chemical plant complex maneuverability and renewable penetration. The values of different parameters used in this scenario are: PHG ramp rate limit Mphg ¼ ð1=48Þ [hr1] and fese ¼ N [hr1] (i.e., no storage). Fig. 14 shows AGD as a function of renewable penetration level for various CPC maneuverability values. These results suggest that: (1) for a given renewable penetration level, as the CPC maneuverability increases (i.e., faster chemical plant complexes), the average generation deficit decreases; (2) for a given CPC maneuverability, as the renewable penetration increases, the average generation deficit increases, hence progressively deviating from the required generation profile. In other words, faster chemical plant complexes facilitate achieving the required generation profile under variable renewable generation. Fig. 15 shows a seven day time snapshot of achievable generation assuming three different values for CPC maneuverability (i.e., Mcpc ¼ 0:36 [hr1], 0.72 [hr1], and 1.8 [hr1]) at two different renewable penetration values (i.e., 20% and 40%). These results suggest that a solution to accommodate more renewable energy is to integrate with relatively faster CPC. As mentioned Section 4.5, the CPC maneuverability can be modified by introducing an AHG or/and (heat transfer element) HTE. 5.4. Effect of renewable penetration on chemical plant complex required maneuverability at different energy accommodations The effect that renewable penetration has on the required CPC maneuverability is evaluated at different energy accommodation levels. At their corresponding scenarios, the reported values for required CPC maneuverability assure that there is no high frequency heat rejection experienced at the CPC. Assuming that the oscillations characterizing the incoming steam can be decomposed on a number of frequency components, a high frequency heat rejection is defined here as the energy carried by heat components oscillating at high frequencies that does not directly translate into

Renewable Penetration =20% 150

Achievable Generation (MW)

100

50

Required Generation 0

M

=0.36 hr−1

−1

Mcpc=0.72 hr−1

cpc

Mcpc=1.8 hr

Renewable Penetration =40% 150

100

50

0 50

51

52

53

54

55

56

Days Fig. 15. Effect of renewable penetration on achievable generation at various chemical plant maneuverability values as function of time.

57

14

H.E. Garcia et al. / Energy 52 (2013) 1e16

completeness, the effectiveness of EA (energy accommodation) in handling the renewable variability is discussed here, but is not pursued any further in the study.

Energy Accommodation [%]

0.4

−1

CPC required maneuverability (hr )

0.45 100%

0.35 95% 0.3

5.5. Effect of chemical plant complex maneuverability and renewable penetration on high frequency heat rejection

More process heat venting

0.25

90%

0.2 0.15

85%

0.1 80% 0.05 0

0

5

10

15

20

25

30

35

40

Renewable penetration level (%) Fig. 16. Effect of renewable penetration on chemical plant required maneuverability at various energy accommodation levels.

useful work (such as into producing chemical products). The values of other parameters used in this scenario are as follows: PHG ramp rate limit Mphg ¼ ð1=48Þ [hr1]. In particular, Fig. 16 shows the required CPC maneuverability as a function of renewable penetration level and energy accommodation. These results suggest that: (1) for a given renewable penetration level, as the energy accommodation increases, the CPC maneuverability needs to be increased in order to meet the required generation profile, and (2) for a given energy accommodation level, as the renewable penetration increases, the CPC maneuverability needs to be increased in order to meet the required generation profile. Venting steam to the atmosphere is unacceptable under normal conditions as it is a wasteful approach for renewable variability accommodation. It is included to ensure that the model reveals the hopefully infrequent consequences of inadequate dynamic responses. In actual operation, it would be done for safety purpose when unforeseen oversupply of steam happens. For sake of

8

7

x 10

−1

CPC high frequency heat rejection

CPC Maneuverability [hr ]

0.36

6

5

Faster chemical plant complex

0.72

4

3 1.8 2 3.6 1 7.2 0

unlimited 0

5

10

15

20

25

30

35

40

Renewable penetration (%) Fig. 17. Effect of renewable penetration on chemical plant high frequency heat rejection at various chemical plant complex maneuverability values.

The effect of renewable penetration and chemical plant required maneuverability on CPC high frequency heat rejection are considered, assuming a value for PHG ramp rate limit equal to Mphg ¼ ð1=48Þ [hr1]. In particular, Fig. 17 shows the effect of renewable penetration on chemical plant high frequency heat rejection at various CPC maneuverability values. It can be seen that at a given renewable penetration level, as the maneuverability of chemical plant increases, the high frequency heat loss rejection decreases. Furthermore, for a given CPC maneuverability value, as the renewable penetration level increases, the high frequency heat rejection increases as well.

6. Conclusions Dynamic analysis of HES was carried out to understand various dynamic challenges and opportunities that may arise from accommodating ever increasing levels of renewable penetration. The dynamic analysis presented shows that direct integration of variable renewable generation poses fundamental technical challenges because of its high variability, unpredictability, and nondispatchability. Furthermore, the analysis shows that there are several strategies that can be utilized to meet required generation profiles while integrating renewable energy, including facilitating flexible operation, utilizing storage, designing primary heat generation with better maneuverability, and devising chemical production processes that can accept high variability in heat input. The actual HES solution for a particular application would be an optimized combination of all these strategies in order to minimize cost and environmental impact, while improving profitability and satisfying operational constraints. As a part of on-going work, a more detailed dynamic model considering thermo-hydraulics and electrical dynamics is being developed. This HES configuration tightly couples a nuclear reactor, a steam power plant (including boilers, pumps, and steam turbines/ generators), a wind power complex (including a number of wind turbines and associated electrical devices), a power grid, and a chemical plant, all interconnected via a steam piping network consisting of multiple pipes, steam headers, and boilers. This will provide more insight into dynamic operation of an HES to directly mitigate the variability of wind generation, thereby imposing on the grid less variability from the combined generation. Using this detailed dynamic model, the suitability of HES for grid connectivity and stabilization under high renewable energy penetration will be also investigated. Optimized control strategies for reactive power management that drives voltage regulation decisions on the grid is a matter of overall grid operation outside the scope of this analysis of a modest micro-grid. The use of detailed dynamic grid dynamics would help in evaluating power fluctuation levels and devising optimized control strategies for reactive power management that drives voltage regulation decisions on the grid. The estimated level of power fluctuations on a grid due to HES can then be compared with a criterion or figure-of-merit obtained from grid operators to demonstrate effectiveness of HES in mitigating variability-induced grid stability issues.

H.E. Garcia et al. / Energy 52 (2013) 1e16

Acknowledgment The authors would like to acknowledge the assistance of Dr. Thomas Baldwin in providing the dynamic model for steam to electrical generation and of Mr. Rick Wood in solving energy balance calculations. The authors would also like to acknowledge Dr. Richard Boardman and Dr. Paul Meakin in reviewing an earlier version of this manuscript and the anonymous reviewers for their constructive and thoughtful comments that helped to improve and clarify it. The research reported here was supported by the Energy Security Initiative (ESI) at Idaho National Laboratory (INL) under the U.S. Department of Energy contract DE-AC0705ID14517. References [1] Department of Energy. Report on the first quadrennial technology review, Tech. rep. Department of Energy. http://energy.gov/sites/prod/files/QTR_ report.pdf; Sep 2011 [accessed 30.01.13]. [2] Cherry R, Aumeier S, Boardman R. Large hybrid energy systems for making low CO2 load-following power and synthetic fuel. Energy and Environmental Science. [3] Cherry R, Boardman R. Using excess electric generation capacity to make synthetic fuels at the national scale. Energy Policy. [4] Cherry R, Garcia H. Multiple hybrid energy systems for load-following and offsetting renewable generations variability. Energy. [5] Perry S, Klemes J, Bulatov I. Integrating waste and renewable energy to reduce the carbon footprint of locally integrated energy sectors. Energy 2008;33(10): 1489e97. [6] Dali M, Belhadj J, Roboam X. Hybrid solar-wind system with battery storage operating in grid-connected and standalone mode: control and energy management e experimental investigation. Energy 2010;35(6):2587e95. [7] Abdullah M, Yung V, Anyi M, Othman A, Hamid KA, Tarawe J. Review and comparison study of hybrid diesel/solar/hydro/fuel cell energy schemes for a rural ICT telecenter. Energy 2010;35(2):639e46. [8] Ibrahim H, Youns R, Basbous T, Ilinca A, Dimitrova M. Optimization of diesel engine performances for a hybrid winddiesel system with compressed air energy storage. Energy 2011;36(5):3079e91. [9] Lau K, Yousof M, Arshad S, Anwari M, Yatim A. Performance analysis of hybrid photovoltaic/diesel energy system under malaysian conditions. Energy 2010; 35(8):3245e55. [10] Reichling J, Kulacki F. Utility scale hybrid windsolar thermal electrical generation: a case study for minnesota. Energy 2008;33(4):626e38. [11] Erdil E, Ilkan M, Egelioglu F. An experimental study on energy generation with a photovoltaic (PV)solar thermal hybrid system. Energy 2008;33(8):1241e5. [12] Rehman S, Al-Hadhrami LM. Study of a solar pvdieselbattery hybrid power system for a remotely located population near Rafha, Saudi Arabia. Energy 2010;35(12):4986e95. [13] Nandi SK, Ghosh HR. Prospect of wind-PV-battery hybrid power system as an alternative to grid extension in Bangladesh. Energy 2010;35(7):3040e7. [14] Traverso A, Magistri L, Massardo A. Turbomachinery for the air management and energy recovery in fuel cell gas turbine hybrid systems. Energy 2010; 35(2):764e77. [15] Carton J, Olabi A. Wind/hydrogen hybrid systems: opportunity for Irelands wind resource to provide consistent sustainable energy supply. Energy 2010; 35(12):4536e44.   [16] Bakic V, Pezo M, Stevanovic Zana, Zivkovic M, Grubor B. Dynamical simulation of PV/wind hybrid energy conversion system. Energy 2011;0. [17] Borba BSM, Szklo A, Schaeffer R. Plug-in hybrid electric vehicles as a way to maximize the integration of variable renewable energy in power systems: the case of wind generation in northeastern Brazil. Energy 2012;37(1):469e81. [18] Cherif H, Belhadj J. Large-scale time evaluation for energy estimation of standalone hybrid photovoltaic wind system feeding a reverse osmosis desalination unit. Energy 2011;36(10):6058e67. [19] Manolakos D, Papadakis G, Papantonis D, Kyritsis S. A simulation-optimisation programme for designing hybrid energy systems for supplying electricity and fresh water through desalination to remote areas: case study: the Merssini village, Donoussa island, Aegean Sea, Greece. Energy 2001;26(7):679e704. [20] Kamel S, Dahl C. The economics of hybrid power systems for sustainable desert agriculture in Egypt. Energy 2005;30(8):1271e81. [21] Atwater T, Cygan P, Leung F. Man portable power needs of the 21st century: I. Application for dismounted soldier. II. Enhanced capabilities through use of hybrid power sources. Journal of Power Sources 2000;91(1):27e36. [22] Jarvis L, Cygan P, Roberts M. Hybrid power source for manportable applications. IEEE Aerospace and Electronic Systems Magazine 2003;18(1):13e6. [23] Abdalla T, Ducey R, Balog R, Feickert C, Weaver W, Abbas A, et al. Control dynamics of adaptive and scalable power and energy systems for military micro grids, Tech. rep., DTIC Document: ERDC/CERL TR-06e35. 2006. [24] Hartranft T. Sustainable energy for deployed military bases. In: 2nd International Conference on Energy Sustainability, Jacksonville, Florida; 2008.

15

[25] Hoover C, Sharma R, Watson B, Charles S, Shah A, Patel C, et al. Sustainable IT ecosystems: enabling next-generation cities, Tech. rep., HP Laboratories Technical Report, HPL-2010-73, http://www.hpl.hp.com/techreports/2010/ HPL-2010-73.html; 2010 [accessed 30.01.13]. [26] Mouawad J. The newest hybrid model. In: the New York Time [accessed 6th July], http://www.nytimes.com/2010/03/05/business/05solar.html; 4 March 2010. [27] Clean energy for the next generation: Martin Next Generation Solar Energy Center, Florida Power and Light Company; http://www.fpl.com/environment/ solar/martin.shtml, [accessed 06.07.11]. [28] Colorado integrated solar project, Xcel Energy. http://www.xcelenergy.com/ Environment/Renewable_Energy/Solar/Colorado_Integrated_Solar_Project, [accessed 06.07.11]. [29] Flexefficiency 50 combined cycle power plant, The General Electric Company; http://www.ecomagination.com/technologies/flex-efficiency/, [accessed 06.07.11]. [30] Hittinger E, Whitacre J, Apt J. Compensating for wind variability using colocated natural gas generation and energy storage. Energy Systems 2010;1: 417e39. [31] Weisser D, Garcia RS. Instantaneous wind energy penetration in isolated electricity grids: concepts and review. Renewable Energy 2005;30(8):1299e 308. [32] Eriksen P, Ackermann T, Abildgaard H, Smith P, Winter W, Rodriguez Garcia J. System operation with high wind penetration. Power and Energy Magazine, IEEE 2005;3(6):65e74. [33] Hamsic N, Schmelter A, Mohd A, Ortjohann E, Schultze E, Tuckey A, et al. Increasing renewable energy penetration in isolated grids using a flywheel energy storage system. Power Engineering, Energy and Electrical Drives 2007:195e200. POWERENG 2007. International Conference on, 2007. [34] Apt J. The spectrum of power from wind turbines. Journal of Power Sources 2007;169(2):369e74. [35] Hesler S. Mitigating the effects of flexible operation on coal-fired power plants. POWER Magazine http://www.powermag.com/business/3886.html, [accessed 30.09.11]. [36] Amenedo JLR. Spain’s example of solving grid stabilization issues, renewable energy world. http://www.renewableenergyworld.com/rea/news/article/ 2010/10/how-spain-dealt-with-lvrt-problems, [accessed 30.09.11]. [37] Carrasco J, Franquelo L, Bialasiewicz J, Galván E, Guisado R, Prats M, et al. Power-electronic systems for the grid integration of renewable energy sources: a survey. Industrial Electronics, IEEE Transactions on 2006;53(4): 1002e16. [38] Boyle G. Renewable electricity and the grid: the challenge of variability. Earthscan/James & James; 2008. [39] Chen Z, Spooner E. Grid power quality with variable speed wind turbines. Energy Conversion, IEEE Transactions on 2001;16(2):148e54. [40] Greer M. Electricity cost modeling calculations. Academic Press; 2010. [41] Parnandi S. Power market analysis tool for congestion management, Master’s thesis, West Virginia University; 2007. [42] Gul T, Stenzel T. Variability of wind power and other renewables: management options and strategies, Tech. rep., International Energy Agency. http:// www.uwig.org/IEA_Report_on_variability.pdf; 2009 [accessed 30.01.13]. [43] GE Energy. Heavy duty gas turbine products, GE energy e F class. http://www. ge-energy.com/products_and_services/index.jsp, [accessed 06.07.11]. [44] Castronuovo E, Lopes J. On the optimization of the daily operation of a windhydro power plant. Power Systems, IEEE Transactions on 2004;19(3):1599e 606. [45] Angarita J, Usaola J. Combining hydro-generation and wind energy: biddings and operation on electricity spot markets. Electric Power Systems Research 2007;77(5e6):393e400. [46] Greenblatt J, Succar S, Denkenberger D, Williams R, Socolow R. Baseload wind energy: modeling the competition between gas turbines and compressed air energy storage for supplemental generation. Energy Policy 2007; 35(3):1474e92. [47] Ummels B, Pelgrum E, Kling W. Integration of large-scale wind power and use of energy storage in the Netherlands’ electricity supply. Renewable Power Generation, IET 2008;2(1):34e46. [48] Black M, Strbac G. Value of bulk energy storage for managing wind power fluctuations. Energy Conversion, IEEE Transactions on 2007;22(1): 197e205. [49] McDowall J. Integrating energy storage with wind power in weak electricity grids. Journal of Power Sources 2006;162(2):959e64. [50] Lund H. Large-scale integration of wind power into different energy systems. Energy 2005;30(13):2402e12. [51] Nemet A, Kleme JJ, Varbanov PS, Kravanja Z. Methodology for maximising the use of renewables with variable availability. Energy 2012;44(1):29e37. [52] Kang CA, Brandt AR, Durlofsky LJ. Optimal operation of an integrated energy system including fossil fuel power generation, CO2 capture and wind. Energy 2011;36(12):6806e20. [53] Niknam T, Azizipanah-Abarghooee R, Roosta A, Amiri B. A new multiobjective reserve constrained combined heat and power dynamic economic emission dispatch. Energy 2012;42(1):530e45. [54] Rajkumar R, Ramachandaramurthy V, Yong B, Chia D. Techno-economical optimization of hybrid pv/wind/battery system using neuro-fuzzy. Energy 2011;36(8):5148e53.

16

H.E. Garcia et al. / Energy 52 (2013) 1e16

[55] Aumeier S, Cherry R, Boardman R, Smith J. Nuclear hybrid energy systems: Imperatives, prospects, and challenges. In: 2nd International Conference on Asian Nuclear Prospect (ANUP 2010), Nagoya, Japan; 2010. [56] Lei M, Shiyan L, Chuanwen J, Hongling L, Yan Z. A review on the forecasting of wind speed and generated power. Renewable and Sustainable Energy Reviews 2009;13(4):915e20. [57] Daren Y, Zhiqiang X. Nonlinear coordinated control of drum boiler power unit based on feedback linearization. IEEE Transactions on Energy Conversion 2005;20(1):204e10. [58] De Mello F. Dynamic models for fossil fueled steam units in power system studies. IEEE Transaction on Power Systems 1991;6(2):753e61. [59] Byerly R. Dynamic models for steam and hydro turbines in power system studies. IEEE Transactions on Power Apparatus and Systems 1973;(6):1904e15. [60] Davis C, Oh C, Barner R, Sherman S, Wilson D. Thermal-hydraulic analyses of heat transfer fluid requirements and characteristics for coupling a hydrogen production plant to a high-temperature nuclear reactor, INL/ET-05e00453, June. [61] Vilim R. Alternate VHTR/HTE interface for mitigating tritium. Tech. rep. Argonne National Laboratory (ANL). http://www.osti.gov/energycitations/ product.biblio.jsp?osti_id¼951247; 2009 [accessed 30.01.13].

[62] Vilim R. NGNP/HTE full-power operation at reduced high-temperature heat exchanger temperatures. Tech. rep. Argonne National Laboratory (ANL). http:// www.osti.gov/bridge/product.biblio.jsp?osti_id¼951249; 2009 [accessed 30.01.13]. [63] Khare Y, Singh Y. PID control of heat exchanger system. International Journal of Computer Applications IJCA 2010;8(6):24e8. [64] Niwa T, Nagaoka N, Mori N, Ametani A, Umeda S. A control method of charging and discharging lithium-ion battery to prolong its lifetime in power compensator for dc railway system. In: 43rd Int. Universities Power Engineering Conference, 2008. UPEC 2008, Nagoya, Japan; 2008. p. 1e5. [65] China claims worlds largest battery storage station. Greenbang: the Smart technology network [cited January 30, 2012]. http://www.greenbang.com/ china-claims-worlds-largest-battery-storage-station_21041.html. [66] Khajavi M, Menhaj M, Suratgar A. A neural network controller for load following operation of nuclear reactors. Annals of Nuclear Energy 2002;29(6): 751e60. [67] Boroushaki M, Ghofrani M, Lucas C, Yazdanpanah M. An intelligent nuclear reactor core controller for load following operations, using recurrent neural networks and fuzzy systems. Annals of Nuclear Energy 2003;30(1): 63e80.