Multiset-based assessment of vulnerability of energy infrastructures to destructive impacts Igor Sheremet Russian Foundation for Basic Research, Leninskiy Prosp., 32a, Moscow, Russia, 119334 sheremet@rfbr.ru Abstract. This paper is dedicated to the application of the multigrammatical framework to the assessment of vulnerability of energy infrastructures affected by impacts destroying (reducing capabilities of) their facilities (power plants, fuel producing plants, power transmission lines, fuel transporting pipes, as well as networking devices of both electricity and fuel subsystems of an energy infrastructures). A basic graph representation of energy infrastructures is considered, and technique of their multigrammatical representation is introduced. Criterial base for recognition of the energy infrastructures vulnerability, being a generalization of the similar criterial base developed regarding industrial infrastructures is proposed. Techniques of multigrammatical modelling reservation of energy infrastructures and their recovery after impacts is proposed. Directions of future research in this area are announced. Keywords. Energy infrastructure, vulnerability, recovery, resilience, multisets, multiset grammars, filtering unitary multiset grammars. 1. Introduction The multigrammatical framework (MGF), introduced and described in [1-7], is a set of syntactically, semantically and pragmatically interconnected multiset-based knowledge representation models (KRMs) and associated with them algorithmics and implementation techniques, developed and applied to various problems from the systems analysis and operations research areas. The MGF integrates the best features of modern knowledge engineering – first of all, logic and constraint programming [8-13], providing easy and natural accumulation of knowledge bases (KBs) from atomary implications and not less easy and natural KBs’ update - and classical theory of optimization – namely, mathematical programming with it’s refined algorithmics providing fast search of strictly optimal solutions [14-17]. The MGF, in fact, provides natural and easily modified representation of distributed sociotechnological systems (DSTSs) of different classes, as well as representation of the so called resource-based games (RBGs) being a useful and convenient tool for modelling various conflicts between DSTSs and their coalitions [18]. One of the most valuable and actual areas of the MGF application is an assessment of DSTSs’ resilience/vulnerability to various destructive impacts (malfunctions, technogenic catastrophes, natural hazards, acts of terror, mutual sanctions etc.). A unified approach to the solution of this class of problems regarding large- scale industrial systems (ISs) was described in [3, 4, 7], whilst techniques of the MGF application to an assessment of resilience of modern intelligent transport systems – in [19]. However, a background of all modern DSTSs is an energy infrastructure (EI), providing production and delivery necessary amounts of electric power _____________ Copyright © 2021 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). and fuel to various stationary and mobile consumers, including industrial facilities, living houses, transportation vehicles etc [20-22]. This paper is dedicated namely to the application of the MGF to some considered from the substantial and mathematical points of view in [22-27] actual tasks from the area of resilience of energy infrastructures. Amounts of electric power (EP) to be delivered by an EI on demand of external customers at some predefined period of time in a general case are restricted by amounts of primary resources – crude oil, natural gas, and other possible energy carriers (ECs) – available for EP generation, as well as by limited bandwidths of links forming electric grids and fuel pipelines. A problem in question is, given a demand of costumers, i.e. amounts of power and fuel to be consumed by them during a considered time period (this demand will be named also an order), an EI segment, including fuel producing and power generating facilities, links providing power transmission and fuel transfer through distributed areas, as well as terminal units delivering fuel and power to their consumers, primary resources available for power generation, a destructive impact, eliminating some part of a considered EI segment and the aforementioned resources, to assess whether a part of a considered segment and resources, remained after an impact, would be capable to produce and deliver amounts of power and fuel necessary to consumers (in other words, to complete an order). If so, then an EI will be named resilient to this impact. Otherwise an EI will be named vulnerable to it. The objective of this paper is to develop a criterial base providing the assessment of EIs vulnerability to destructive impacts. Everywhere below in this paper we shall consider an EI as a closed system, which operate without direct application of any external resources or their application for replenishment of EI own (internal) resources spent whilst order completion. A content of this paper is as follows. A basic graph representation of EIs is introduced and discussed in the Section 2. Filtering unitary multiset grammars being a basic tool for consideration and solution of the problem in question are described in the Section 3. A multigrammatical representation of energy infrastructures is proposed in the Section 4 whilst criteria of vulnerability of energy infrastructures to destructive impacts – in the Section 5. Modelling reservation of EIs and their recovery after impacts is considered in the Section 6. A Conclusion is dedicated to the future directions of the MGF development and it’s application to various issues concerning resilience of critical infrastructures and key resources. 2. Basic graph representation of energy infrastructures An energy infrastructure is usually considered consisting of two strongly interconnected and mutually supplying segments producing fuel and electricity [20-22]. An electricity infrastructure (ElcI) in the most general case contains generation facilities (power plants, PPs), power transforming-distributing substations (PTDSs), and power terminal units (PTUs), delivering electric power to it’s consumers. All these elements are connected by links, named power transmission lines (PTLs), each such line having it’s own technical parameters (voltage, length, power losses during transmission etc.), and are joined to electric grids, which , in fact, in aggregate form ElcI [21-23]. A fuel infrastructure (FI) [24, 25, 28, 29], similarly to an ElcI, includes fuel producing plants (FPPs), working out fuel from some primary energy carriers (PECs), and fuel distribution stations (FDSs), as well as fuel terminal units (FTUs). All these elements are connected by pipes, which, in a general case, as PTLs, have individual technical parameters (diameter, length, pressure, amounts of EP consumed, fuel losses during transfer etc.). Fuel produced by FPPs is used by power plants and other consumers. To limit a complexity of consideration here, we shall not expand a FI down to production crude oil and natural gas from oil and gas fields and their transportation via oil and gas pipelines to FPPs; we shall assume that certain amounts of primary energy carriers (PECs), used for fuel production, are accumulated at fuel storages (FSs) collocated with FPPs, and these amounts are a part of a resource base (RB) of an EI. ElcI and FI are joined with one another by terminal units: any element of an FI consumes an electric power delivered to it by some PTU, whilst any PP is operating due to a FTUs delivering fuels needed for power generation (in a general case there may be several energy carriers utilized by a single power plant). Also there are PTUs and FTUs delivering power and fuels to external consumers. Regarding a considered time period (hour, day etc.), any FPP may produce certain amounts of various fuels, as well as any PP may produce certain amounts of EP with various technical parameters. Any output of any element of EI is assumed consistent with a link transferring resource from it to another element, which input, in turn, is assumed consistent with the aforementioned link which is an incoming for this another element and thus delivering to it the aforementioned resource. This overlapping of EI elements and boundary points of EI links is a background for modelling a circulation of an EP and fuel via EI. Any link has a limited bandwidth (or throughput capacity) as an integral technical parameter, determining maximal amount of power (if it is a PTL) or fuel (if it is a pipe) which may be transmitted (transferred) via this link during a considered time period. Also, as it was mentioned above, there are some power losses occurring during it’s transmission via a PTL; similar losses of fuel are inherent to fuel transferring pipes. So both electricity and fuel infrastructures have a tree-like concentric topology and, based on the above, an EI may be represented by an weighted oriented graph with nodes corresponding to EI elements, and marked edges corresponding to EI links. This graph, in turn, in the algebraic representation is a ternary relation 𝐺  𝐴 × 𝐴 × 𝑵, where 𝐴 is a set of EI elements (PPs, PTDSs, PTUs, FPPs, FDSs, FTUs, FSs), and 𝑵 is a set of positive rational numbers representing bandwidths of EI links (PTLs and pipes). So < 𝑎, 𝑎′, 𝑛 >∈ 𝐺 means that an element 𝑎 is capable to transmit (transfer) to an element 𝑎′ amount of resource (EP or fuel) by link (PTL or pipe) < 𝑎, 𝑎′ > no more than 𝑛 units (kilowatt∙hours in the case of EP, and barrels, cubic meters, kilograms, tons etc. in the case of various fuels) during a considered time period. There may be the only triple < 𝑎, 𝑎′, 𝑛 > ∈ 𝐺 for any link < 𝑎, 𝑎′ > , i.e. a link has the only bandwidth (throughput capacity). A destructive impact, which in a general case is distributed, may eliminate some elements or/and links of an EI as well as some amounts of resources stored at an EI resource base; naturally, an impact may be represented by some subset of nodes and edges eliminated from an initial graph 𝐺. Let us illustrate the said by an example. Example 1. Consider a small hypothetical segment of some EI including a power plant, two power transformation-distribution stations, seven power terminal units, a fuel producing plant, a fuel storage, two fuel distribution stations, and three fuel terminal units (figure 1(a)). (Sequential numbers of FDSs and FTUs, as well as names of fuel storage and fuel producing plant are denoted by bold symbols). There are also three external power customers. Generated power from a PP is delivered to both PTDSs, the first of which (enumerated “1”) delivers received power to four PTUs ( “1”, “2”, “3” and “7”), and the second (“2”) delivers accepted power to five PTUs (“4”, “5”, “6”, “8” and “9”). PTUs deliver power to the following elements of the EI: PTU “1” – to FDS “2”, PTU “2” – to FDS “1”, PTU “3” – to FPP, PTU “4” – to FS, PTU “5” – to the power customer “1”, “6” – to the power customer “2”, “7” – to the power customer “3”. In turn, elements of FI by consumption of electric power deliver fuel as follows: FS – to FPP, FPP – to FDS “1”, FDS “1” – to FDS “2”, FDS “2” – to FTU “1” collocated with power plant. FTUs “2” and “3”, which both receive fuel from FDS “1”, deliver fuel to fuel customers “1” and “2” respectively. These FTUs are provided by electric power from PTUs “8” and “9”, receiving power from the PTDS “2”. An algebraic representation of the considered graph, including bandwidths (throughput capacities) is contained in the table 1. The impact destroys the PTDS “1”, PTUs “5” and “7”, as well as the FDS “2”. Along with these destructions the impact reduces bandwidth of the link between the PTDS “2” and the PTU “6” from 300 kWh to 100 kWh. The resulting graph of the affected EI is represented at figure 1(b). ∎ Having this basic graph representation of EIs we may move to the MGF application to the assessment of resilience/vulnerability of EIs. To introduce proposed a criterial base for this assessment let us remind some necessary notions and denotations concerning syntax and semantics of filtering unitary multiset grammars (FUMGs) being a simplest MGF tool for formalizing and solution of many actual tasks from the applied systems analysis and operations research areas. a) b) Figure 1. Graph representation of a segment of an energy infrastructure a) initial state, b) state after impact. Table 1. An algebraic representation of the graph № Source point Receiver point Channel upper threshold values of bandwidths (throughput capacities) 1 PP PTDS1 1000 kWh 2 PP PTDS2 1100 kWh 3 PTDS1 PTU1 200 kWh 4 PTDS1 PTU2 300 kWh 5 PTDS1 PTU3 400 kWh 6 PTDS1 PTU7 100 kWh 7 PTDS2 PTU4 200 kWh 8 PTDS2 PTU5 300 kWh 9 PTDS2 PTU6 300 kWh № Source point Receiver point Channel upper threshold values of bandwidths (throughput capacities) 10 PTDS2 PTU8 200 kWh 11 PTDS2 PTU9 100 kWh 12 PTU5 EPC1 300 kWh 13 PTU6 EPC2 300 kWh 14 PTU7 EPC3 100 kWh 15 FS FPP 200 tons of crude oil 16 FPP FDS1 200 tons of the fuel 17 FDS1 FDS2 100 tons of the fuel 18 FDS1 FTU2 50 tons of the fuel 19 FDS1 FTU3 50 tons of the fuel 20 FDS2 FTU1 100 tons of the fuel 21 FTU1 PP 100 tons of the fuel 22 FTU2 FC1 50 tons of the fuel 23 FTU3 FC2 50 tons of the fuel 3. Filtering unitary multiset grammars Following [1, 2, 3], we shall define a multiset grammar (multigrammar, MG) as a couple 𝑆 =< 𝑣0 , 𝑅 >, (1) where a multiset (MS) 𝑣0 = {𝑛1 ∙ 𝑎1 , … , 𝑛𝑚 ∙ 𝑎𝑚 }, (2) is called a kernel, and 𝑅, called a scheme, is a finite set of rules which are applied for generation new multisets from already generated. (Everywhere below objects are denoted 𝑎, 𝑎𝑖 , 𝑎𝑖𝑗 , whilst their multiplicities being positive rational numbers – as 𝑛𝑖 , 𝑚𝑖 , 𝑛𝑖𝑗 etc.; a construction 𝑛𝑖 ∙ 𝑎𝑖 representing collection of 𝑛𝑖 objects 𝑎𝑖 is called a multiobject). A rule has a form 𝑣 → 𝑣′, (3) where 𝑣 and 𝑣 ′ , called respectively the left part and the right part of a rule, are multisets, and 𝑣 ≠ {∅}. By 𝐴𝑠 we shall designate below a set of all objects having place in rules entering a scheme 𝑅 of an MG 𝑆. The semantics of a rule is defined on the background of the relation of inclusion on multisets, denoted  , and operations of addition and subtraction of multisets, denoted respectively + and - . Let 𝑣̅ be a multiset. A rule (3) is applicable to 𝑣̅ , if 𝑣̅  𝑣, (4) and a result of an application is a multiset 𝑣̅ ′ = 𝑣̅ - 𝑣 + 𝑣′, (5) i.e. if 𝑣̅ includes 𝑣, then 𝑣 is replaced by 𝑣′. This operation is called a generation step, providing a generation an MS 𝑣̅ ′ from an MS 𝑣̅ by application a rule 𝑟 ∈ 𝑅, that is denoted as 𝑟 𝑣̅  𝑣̅ ′ , (6) whilst a fact, that an MS 𝑣̅ ′ is generated from an MS 𝑣̅ by any (including empty) sequence of generation steps, called a generation chain, is recorded as 𝑅 𝑣̅  𝑣̅ ′ , (7) or, if the only MG is considered, then, as in the classic string-operating grammars [28, 29], ∗ 𝑣̅  𝑣̅ ′ . (8) + ∗ If a generation chain is non-empty, a denotation  instead of  is used. A set of multisets (SMS), generated by an MG 𝑆 =< 𝑣0 , 𝑅 >, is denoted 𝑉𝑠 and is defined as follows: 𝑅 𝑉𝑠 = {𝑣 | 𝑣0  𝑣}.( (9) An MS 𝑣 is called a terminal multiset (TMS), if there is no one rule 𝑟 𝑅 which may be applied to 𝑣. A set of terminal sets (STMS) will be denoted 𝑉̅𝑠 . Obviously, 𝑉̅𝑠  𝑉𝑠 . Unitary multiset grammars (UMGs) are a simplified version of a partial case of MGs, called context-free multigrammars. A scheme of an UMG is a set of unitary rules (URs), where an UR is recorded as 𝑎𝑖0 → 𝑚1 ∙ 𝑎𝑖1 , … , 𝑚𝑘 ∙ 𝑎𝑖𝑘 , (10) that is equivalent to {1 ∙ 𝑎𝑖0 } → {𝑚1 ∙ 𝑎𝑖1 , … , 𝑚𝑘 ∙ 𝑎𝑖𝑘 }. (11) The left part of an UR being an object 𝑎𝑖0 is called it’s header, whilst the right one – it’s body. A set of non- terminal objects, each being a header of at least one UR, is denoted 𝐴𝑆𝑁 ; and a set of all other objects, presenting only in bodies of URs and called terminal , is denoted 𝐴𝑆 : 𝐴𝑆 = 𝐴𝑆𝑁 ∪ 𝐴𝑆 , (12) 𝐴𝑆𝑁 ∩ 𝐴𝑆 = {∅}, (13) 𝐴𝑆  𝑽+ , (14) where 𝑽+ is a set of non-empty strings in some primary alphabet 𝑽 used for construction of objects’ names. Everywhere below bold letters in objects’ names will be assumed entering an alphabet 𝑽, and bold letters “(“, ”)”, ”[“, ”]”, ”:” will be delimiters entering 𝑽 and used for construction of object names entering a set 𝐴𝑆 . UMGs may be classified by number of URs having the same header. If an UMG 𝑆 =< 𝑣0 , 𝑅 > is such that in a scheme 𝑅 there exists at least one non-terminal object being of header of 𝑚 > 1 URs, then this UMG is named alternating; otherwise, i.e. if any non-terminal object is a header of the only one UR, then this UMG is named non-alternating. Evidently, if an UMG 𝑆 is non-alternating, then it defines a one-element STMS, i.e. |𝑉 ̅𝑠 |=1. If ̅ an UMG 𝑆 is alternating, then in a general case it defines a set containing no less than one TMS, i.e. |𝑉𝑠 |≥1. Also UMGs may be cyclic or non-cyclic. An UMG 𝑆 = 〈𝑣0 , 𝑅〉 will be called cyclic, if there exists a ∗ + generation chain 𝑣0 ⇒ 𝑣 ⇒ 𝑣′ such that 𝑣 ⊆ 𝑣′, or, just the same, 𝑣 ′ = 𝑣 + 𝛥𝑣, where 𝛥𝑣 ⊇ {∅}. As may be seen, a cyclic UMG in a general case, when 𝛥𝑣 ⊇ {∅} but 𝛥𝑣 ≠ {∅}, defines an infinite STMS 𝑉̅𝑠 . All UMGs which are not cyclic, are named acyclic. Any acyclic UMG 𝑆 = 〈𝑣0 , 𝑅〉 defines a finite STMS 𝑉̅𝑠 . By finite or infinite number of elements of an STMS defined by an UMG 𝑆 = 〈𝑣0 , 𝑅〉 it may finitary (in this case |𝑉 ̅𝑠 |< ∞) or infinitary (in this case |𝑉 ̅𝑠 |= ∞). As it is known from [1, 2], any infinitary UMG is obligatory cyclic, while any finitary UMG is acyclic. There exist cyclic UMGs being finitary. Alternating UMGs are a standard tool for representation of alternative structures of complex (composite) objects or ways of solution of some task. This class of UMGs is for a long time used for modelling industrial systems and infrastructures [1-7]. From the other side, cyclic UMGs may be applied to a description of interconnected processes and critical infrastructures with mutual resource exchange; for example, a fuel infrastructure produces a fuel which is consumed by an electricity infrastructure, in turn, providing operation of facilities of a FI. Such UMGs will be applied below in this paper for representation and consideration of energy infrastructures. Example 2. Consider the UMG 𝑆 =< 𝑣0 , 𝑅 >, where 𝑣0 = {2 ∙ (𝒂𝒖𝒕𝒐)}, and the scheme 𝑅 contains three unitary rules 𝑟1 , 𝑟2 and 𝑟3 : 𝑟1 : (𝒂𝒖𝒕𝒐) → 1 ∙ (𝒇𝒓𝒂𝒎𝒆), 1 ∙ (𝒆𝒏𝒈𝒊𝒏𝒆), 4 ∙ (𝒘𝒉𝒆𝒆𝒍), 4 ∙ (𝒅𝒐𝒐𝒓), (15) 400 ∙ (𝒌𝑾𝒉), 50 ∙ (𝒎𝒏𝒕: 𝒂𝒖𝒕𝒐𝒔 𝑨𝑳); 𝑟2 : (𝒆𝒏𝒈𝒊𝒏𝒆) → 1 ∙ (𝒎𝒐𝒕𝒐𝒓), 1 ∙ (𝒇𝒖𝒆𝒍 𝒕𝒂𝒏𝒌), (16) 100 ∙ (𝒌𝑾𝒉), 60 ∙ (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟏𝑨𝑳); 𝑟3 : (𝒆𝒏𝒈𝒊𝒏𝒆) → 1 ∙ (𝒎𝒐𝒕𝒐𝒓), 1 ∙ (𝒇𝒖𝒆𝒍 𝒕𝒂𝒏𝒌), (17) 80 ∙ (𝒌𝑾𝒉), 70 ∙ (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟐𝑨𝑳). The kernel of this UMG represents the order, which objective is to obtain two autos, whilst the scheme represents the so called manufacturing technological base of some industrial facility capable to complete such orders. The UR 𝑟1 represents the structure of auto, which consists of frame, engine, 4 wheels and 4 doors, as well as resources necessary for assembling this auto: 400 kilowatt∙hours of electric power and 50 minutes of operation of autos assembling line (AL). The URs 𝑟2 and 𝑟3 represent structure of engine (motor and fuel tank), and two alternative ways of it’s manufacturing by two engines assembling lines, the first consuming 100 kilowatt∙hours and 60 minutes, and the second – 80 kilowatt∙hours and 70 minutes for one engine. According to the semantics of UMGs, 𝑉̅𝑠 = {𝑣2,0 , 𝑣0,2 , 𝑣1,1 }, where 𝑣2,0 represents total of resources necessary for manufacturing both autos by the first way (involving the first engines AL), 𝑣0,2 – similar value when both engines are assembled by the second such AL, and 𝑣1,1 – when engines are assembled in parallel by separate ALs. Evidently, 𝑣2,0 = 𝒗 + {1000 ∙ (𝒌𝑾𝒉), 120 ∙ (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟏𝑨𝑳)}, (18) 𝑣0,2 = 𝒗 + {960 ∙ (𝒌𝑾𝒉), 140 ∙ (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟐𝑨𝑳)}, (19) 𝑣1,1 = 𝒗 + {980 ∙ (𝒌𝑾𝒉), 60 ∙ (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟏𝑨𝑳), 70 ∙ (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟐𝑨𝑳)}, (20) where 2 ∙ (𝒇𝒓𝒂𝒎𝒆), 2 ∙ (𝒆𝒏𝒈𝒊𝒏𝒆), 8 ∙ (𝒘𝒉𝒆𝒆𝒍), 8 ∙ (𝒅𝒐𝒐𝒓), 2 ∙ (𝒎𝒐𝒕𝒐𝒓), 𝒗={ }. (21) 2 ∙ (𝒇𝒖𝒆𝒍 𝒕𝒂𝒏𝒌), 100 ∙ (𝒎𝒏𝒕: 𝒂𝒖𝒕𝒐𝒔 𝑨𝑳) ∎ We shall use below filtering unitary multiset grammars (FUMGs) as a basic mathematical tool for representation and solution of tasks in question. According to [1, 2], a FUMG is a triple 𝑆 =< 𝑣0 , 𝑅, 𝐹 >, (22) where an UMG 𝑆′ =< 𝑣0 , 𝑅 > is called a core UMG of a FUMG 𝑆 , and 𝐹 is a filter, i.e. a set of so called boundary and optimizing conditions on multiplicities of objects specified in a filter. A filter provides selection from an STMS, generated by an UMG 𝑆′, terminal multisets satisfying aforementioned conditions. A boundary condition (BCs) is recorded as 𝑎𝜃𝑛 or ′ , where 𝜃 ∈ {≥, >, <, ≤, =, ≠}, whilst an optimizing condition (OC) is recorded as 𝑎 = 𝑜𝑝𝑡, where 𝑜𝑝𝑡 ∈ {𝑚𝑖𝑛, 𝑚𝑎𝑥}. So in a general case 𝐹 = 𝐹≤ ∪ 𝐹𝑜𝑝𝑡 , (23) where 𝐹≤ is a set of BCs, and 𝐹𝑜𝑝𝑡 is a set of OCs. Semantics of filters, in fact, is very similar to semantics of relational query languages if to consider a set 𝑉̅𝑠′ as a specific database (however, infinite in a general case); also, due to application of OCs, filters provide natural representation of various tasks from the area of mathematical programming and, in general, operations research [1, 2]. Formally, semantics of UMGs and FUMGs are interconnected by the following relation: ̅𝑠 = (𝑉̅𝑠′ ↓ 𝐹≤ ) ↓ 𝐹𝑜𝑝𝑡 , 𝑉 (24) where symbol ↓ denotes an operation of filtration: an STMS, generated by an UMG 𝑆′, is filtered by a set of BCs, and then a resulting subset, including TMSs, satisfying all BCs entering 𝐹≤ , is filtered by a set of OCs, so, ̅𝑠 includes TMSs satisfying not only all BCs but also all OCs. finally, 𝑉 Example 3. Let us consider now the FUMG 𝑆 =< 𝑣0 , 𝑅, 𝐹 >, where 𝑣0 and 𝑅 are the same as above, and 𝐹 = { (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟏𝑨𝑳) > 0, (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟐𝑨𝑳) > 0}. (25) From the substantial point of view this filter provides selection of such ways of order completion where no ̅𝑠 = {𝑣1,1 }. If one engines AL is out of operation (both such assembling lines are involved). So, obviously, 𝑉 𝐹 = { (𝒌𝑾𝒉) = 𝑚𝑖𝑛} (26) ̅𝑠 = i.e. such ways of order completion are preferable which consume minimal amount of electric power, then 𝑉 {𝑣0,2 }. In the case 𝐹 = { (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟏𝑨𝑳) > 0, (𝒎𝒏𝒕: 𝒆𝒏𝒈𝒊𝒏𝒆𝒔 𝟐𝑨𝑳) > 0, (𝒌𝑾𝒉) = 𝑚𝑖𝑛}, (27) 𝑉̅𝑠 = {𝑣1,1 } ↓ { (𝒌𝑾𝒉) = 𝑚𝑖𝑛} = {𝑣1,1 }. (28) ∎ According to features of their core UMGs, filtering UMGs may be alternating or non-alternating, cyclic and acyclic, finitary or infinitary. However, due to an application of it’s filter a FUMG, which core UMG is infinitary, may be finitary [1, 2]: a filter may select a finite subset of an infinite STMS defined by a core UMG of an FUMG. Now, at last, we may move directly to the application of FUMGs to the assessment of resilience/vulnerability of energy infrastructures, beginning from a multigrammatical representation of EIs. 4. Basic multigrammatical representation of energy infrastructures Let us begin from an electricity infrastructure. We shall use in URs below names of objects which syntax will be (𝒌𝑾𝒉: 𝑝), where the string 𝒌𝑾𝒉 denotes a measurement unit of EP transmitted via PTLs (kilowatt∙hour), and 𝑝 is a string in an alphabet 𝑽 representing a geographical point, where an element of an ElcI is located (it may be designated by a unique symbolic name associated with specific geographic coordinates in a special database, or directly by these coordinates). So a multiobject 𝑛 ∙ (𝒌𝑾𝒉: 𝑝) represents 𝑛 kilowatts generated or consumed at a point (position, place) 𝑝. Let us begin our consideration from power terminal units. Any PTU in order to deliver one unit of power to a consumer, switched to this PTU, must receive it from a closest PTDS, connected with it by a PTL. So a unitary rule, representing this fragment of an ElcI, would be as follows: (𝒌𝑾𝒉: 𝑝𝑡𝑢) → 𝑛 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠), 𝑛 ∙ [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑢], (29) where 𝑝𝑡𝑢 and 𝑝𝑡𝑑𝑠 are strings, representing locations of, respectively, a PTU and a supplying it PTDS, whilst [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑢] is a string, representing a connecting them PTL. In other words, [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑢] is an object representing a PTL, which start and final points are respectively 𝑝𝑡𝑑𝑠 and 𝑝𝑡𝑢. A value 𝑛 ≥ 1 depends, finally, on amounts of power losses occurring during it’s transmission via a PTL (in the case 𝑛 = 1 there are no any such losses); 𝑛 is a rational number. So a multiobject 𝑛 ∙ [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑢] represents a fact that a considered PTL provides transmission of one kilowatt∙hour to a PTU located at a point 𝑝𝑡𝑢, receiving 𝑛 kilowatt∙hours from a PTDS located at a point 𝑝𝑡𝑑𝑠. (Let us note that the sense of (29) is fully similar to the sense of (10) regarding industrial systems and called a technological interpretation of unitary rules [3, 4, 7], which is illustrated by (15)- (17); namely, to “create” one kilowatt∙hour at a point 𝑝𝑡𝑢 it is necessary to have 𝑛 kilowatt∙hours at a point 𝑝𝑡𝑑𝑠 and also a PTL connecting both points and able to transmit this amount of EP from 𝑝𝑡𝑑𝑠 to 𝑝𝑡𝑢. Similar logics will be applied everywhere above to all components of ElcI and FI). If a PTDS, located at a point 𝑝𝑡𝑑𝑠, is connected to power terminal units, located at points 𝑝𝑡𝑢1 , … , 𝑝𝑡𝑢𝑚 , then this fragment of an ElcI is represented by 𝑚 following unitary rules: (𝒌𝑾𝒉: 𝑝𝑡𝑢1 ) → 𝑛1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠), 𝑛1 ∙ [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑢1 ], … (30) (𝒌𝑾𝒉: 𝑝𝑡𝑢𝑚 ) → 𝑛𝑚 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠), 𝑛𝑚 ∙ [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑢𝑚 ]. Similarly may be represented fragments of an ElcI, consisting of connected PTDSs. In this case a string 𝑝𝑡𝑑𝑠 is a representation of a location of a delivering power transforming-distributing substation, whilst 𝑝𝑡𝑑𝑠1 , … , 𝑝𝑡𝑑𝑠𝑙 – locations of such PTDSs, which are consuming power transformed and transmitted by it: ((𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠1 ) → 𝑛1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠), 𝑛1 ∙ [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑑𝑠1 ], … (31) (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠𝑙 ) → 𝑛𝑙 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠), 𝑛𝑙 ∙ [𝑝𝑡𝑑𝑠, 𝑝𝑡𝑑𝑠𝑙 ]. In such a way all tree-like fragments of an ElcI are represented, until a power plant, producing electric power. Any tree-like fragment of an ElcI, containing some PP and connected with it PTDSs, may be represented by following URs: (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠1 ) → 𝑛1 ∙ (𝒌𝑾𝒉: 𝑝𝑝), 𝑛1 ∙ [𝑝𝑝, 𝑝𝑡𝑑𝑠1 ], … (32) (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠𝑙 ) → 𝑛𝑙 ∙ (𝒌𝑾𝒉: 𝑝𝑝), 𝑛𝑙 ∙ [𝑝𝑝, 𝑝𝑡𝑑𝑠𝑙 ], and, if there are some power terminal units connected to a power plant directly, i.e. without any intermediate PTDSs, then also (𝒌𝑾𝒉: 𝑝𝑡𝑢1 ) → 𝑛1 ∙ (𝒌𝑾𝒉: 𝑝𝑝), 𝑛1 ∙ [𝑝𝑝, 𝑝𝑡𝑢1 ], … (33) (𝒌𝑾𝒉: 𝑝𝑡𝑢𝑚 ) → 𝑛𝑚 ∙ (𝒌𝑾𝒉: 𝑝𝑝), 𝑛𝑚 ∙ [𝑝𝑝, 𝑝𝑡𝑢𝑚 ], where 𝑝𝑝 is a location of a power plant. A power plant, in turn, may be represented by an UR (𝒌𝑾𝒉: 𝑝𝑝) → 𝑛1 ∙ (𝑟𝑒𝑠1 : 𝑝1 ), … , 𝑛𝑘 ∙ (𝑟𝑒𝑠𝑘 : 𝑝𝑘 ), (34) where 𝑛1 , … , 𝑛𝑘 are amounts of resources 𝑟𝑒𝑠1 , … , 𝑟𝑒𝑠𝑘 , which must be delivered to locations 𝑝1 , … , 𝑝𝑘 respectively in order to generate one kilowatt∙hour of electric power at a location 𝑝𝑝, from which, in turn, it may be delivered by PTLs to PTDSs (PTUs), closest to a PP. By this, evidently, 𝑝1 , … , 𝑝𝑘 are locations of terminal units of a fuel infrastructure, which, in turn, delivers the aforementioned resources – most frequently, natural gas and various oil derivatives, which are transferred to power plants by pipelines, as it was described in the Section II. Fuel terminal units, delivering resources to consumers, are represented as headers of unitary rules of the form (𝑟𝑒𝑠: 𝑓𝑡𝑢) → 𝑛 ∙ (𝑟𝑒𝑠: 𝑓𝑑𝑠), 𝑚 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢), 𝑛 ∙ [𝑓𝑑𝑠, 𝑓𝑡𝑢], (35) where multiobject 𝑚 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢) represents a PTU of an electricity infrastructure, located at a point 𝑝𝑡𝑢 and providing operation of an FTU located at a point 𝑓𝑡𝑢 during delivery of one unit of a resource 𝑟𝑒𝑠 from a point 𝑓𝑑𝑠 to a point 𝑓𝑡𝑢. This amount of power is consumed during a resource transfer via a pipe, which start point is 𝑓𝑑𝑠 and final point is 𝑓𝑡𝑢. In a general case, due to losses of fuel during it’s transfer via a pipe, 𝑛 ≥ 1 units of fuel are needed to be delivered to a pump at a start point of this pipe. Distributing facilities (namely, FDSs) of fuel infrastructure may be represented similarly to PTDSs: (𝑟𝑒𝑠: 𝑓𝑑𝑠1 ) → 𝑛1 ∙ (𝑟𝑒𝑠: 𝑓𝑑𝑠), 𝑚1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢1 ), 𝑛1 ∙ [𝑓𝑑𝑠, 𝑓𝑑𝑠1 ], … (36) (𝑟𝑒𝑠: 𝑓𝑑𝑠𝑘 ) → 𝑛𝑘 ∙ (𝑟𝑒𝑠: 𝑓𝑑𝑠), 𝑚𝑘 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢𝑘 ), 𝑛𝑘 ∙ [𝑓𝑑𝑠, 𝑓𝑑𝑠𝑘 ]. (𝑟𝑒𝑠: 𝑓𝑡𝑢1 ) → 𝑛′1 ∙ (𝑟𝑒𝑠: 𝑓𝑑𝑠), 𝑚′1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′1 ), 𝑛′1 ∙ [𝑓𝑑𝑠, 𝑓𝑡𝑢1 ], … (37) (𝑟𝑒𝑠: 𝑓𝑡𝑢𝑙 ) → 𝑛′𝑙 ∙ (𝑟𝑒𝑠: 𝑓𝑑𝑠), 𝑚′𝑙 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′𝑙 ), 𝑛′𝑙 ∙ [𝑓𝑑𝑠, 𝑓𝑡𝑢𝑙 ], that means that delivered resource, incoming to any FDS, is distributed to 𝑘 + 𝑙 pipes by application of corresponding needed amounts of electric power. The first 𝑘 pipes provide fuel transfer to another FDSs whilst the last 𝑙 – to FTUs. As above, [𝑓𝑑𝑠, 𝑓𝑡𝑢𝑖 ], 𝑖 = 1, … , 𝑙, are pipes, which start point is 𝑓𝑑𝑠 and final points are 𝑓𝑡𝑢𝑖 . Similarly, [𝑓𝑑𝑠, 𝑓𝑑𝑠𝑗 ], 𝑗 = 1, … , 𝑘, are pipes, which start point is 𝑓𝑑𝑠 and final points are 𝑓𝑑𝑠𝑗 . Presence of objects (𝒌𝑾𝒉: 𝑝𝑡𝑢𝑖 ) in all unitary rules (36) and objects (𝒌𝑾𝒉: 𝑝𝑡𝑢′𝑗 ) in all unitary rules (37) means that power terminal units, belonging to an electricity infrastructure, would be installed and operate at some predefined points 𝑝𝑡𝑢𝑖 and 𝑝𝑡𝑢′𝑗 respectively to make possible physical contact with FDSs and FTUs and their power supply during transfer of resource 𝑟𝑒𝑠. As it was mentioned above, in a general case every pipe has it’s own technical parameters – finally, it’s own amounts of electric power consumed, i.e. 𝑚𝑖 and 𝑚′𝑗 , as well as losses of a fuel during it’s transfer via this pipe, i.e. 𝑛𝑖 and 𝑛′𝑗 . As it is clear, the described techniques may be applied until places of origination of energy carriers, i.e. fuel production plants, working out pipeline gas and various oil derivatives, used as a fuel by power plants. As it was assumed above, PECs, used for fuel production, are accumulated at fuel storages collocated with FPPs. So operation of any such FPP may be represented as follows: (𝑟𝑒𝑠: 𝑓𝑝𝑝) → 𝑛 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢), 𝑚1 ∙ (𝑟𝑒𝑠1 : 𝑓𝑠1 ), 𝑛1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢1 ), (38) …, 𝑚𝑡 ∙ (𝑟𝑒𝑠𝑡 : 𝑓𝑠𝑡 ), 𝑛𝑡 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢𝑡 ), where 𝑓𝑠1 , … , 𝑓𝑠𝑡 are points, where fuel storages with PECs 𝑟𝑒𝑠1 , … , 𝑟𝑒𝑠𝑡 are located, so namely regarding these places power terminal units would be installed, thus providing relocation of amounts of these PECs necessary to an FPP for production of one unit of fuel 𝑟𝑒𝑠 at a location 𝑓𝑝𝑝. The aforementioned relocation would be possible if needed amounts of electric power, i.e. 𝑛1 ,…, 𝑛𝑡 kilowatt∙hours, would be available at points 𝑝𝑡𝑢1 , … , 𝑝𝑡𝑢𝑡 where respective PTUs are operating. In turn, to produce one unit of a fuel 𝑟𝑒𝑠 an FPP itself would consume 𝑛 kilowatt∙hours from a power terminal unit located at a point 𝑝𝑡𝑢. One more nuance connected with a multigrammatical representation of an energy infrastructures and assessment of their resilience is representation of active states of EI elements. To represent the fact that any producing or transmitting (transferring) facility (PP, FTP, PTDS, FTDS, PTU, FTU) to carry out it’s functions would be in an active state we shall apply techniques proposed and described in [3, 4, 7] regarding industrial systems and based on inclusion to bodies of unitary rules special multiobjects. So in the case of URs (24)-(37) concerning ElcI any unitary rule (𝒌𝑾𝒉: 𝑥) → 𝑋, (39) where 𝑋 is a body of this UR, would be transformed to (𝒌𝑾𝒉: 𝑥) → 𝑋, 1 ∙ (+𝑥), (40) where symbol “+” means that a facility 𝑥 is in an active state and may produce one kilowatt∙hour of an EP. Similarly, unitary rules (35)-(38) concerning FI (𝑟𝑒𝑠: 𝑥) → 𝑋 (41) would be transformed to (𝑟𝑒𝑠: 𝑥) → 𝑋, 1 ∙ (+𝑥). (42) This means that a facility 𝑥 is in an active state and may produce one unit of a resource 𝑟𝑒𝑠. Following [3,4,7], we shall use below the notion “operation cycle of a facility 𝑥” (for short OCF), understanding it as an action performed by a facility to produce one unit of EP, fuel or some other resource. A set (not obligatory a sequence) of 𝑙 such OCFs inside a considered time period of an EI operation has an evident representation by a multiobject 𝑙 ∙ (+𝑥). We shall denote a set of unitary rules representing ElcI, FI and their interconnections, as described above, by 𝑅𝐸 . Let us illustrate techniques of construction such set given a graph representation of an EI. Example 4. Consider the EI segment represented by the graph at Fig. 1a and Table 1. It may be also represented as a following set of unitary rules: (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟏) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒑), 1 ∙ [𝑷𝒑, 𝑷𝒕𝒅𝒔𝟏] , 1 ∙ (+𝑷𝒕𝒅𝒔𝟏) ; (43) (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟐) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒑), 1 ∙ [𝑷𝒑, 𝑷𝒕𝒅𝒔𝟐] , 1 ∙ (+𝑷𝒕𝒅𝒔𝟐) ; (44) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟏) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟏), 1 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟏] , 1 ∙ (+𝑷𝒕𝒖𝟏) ; (45) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟐) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟏), 1 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟐] , 1 ∙ (+𝑷𝒕𝒖𝟐) ; (46) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟑) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟏), 1 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟑] , 1 ∙ (+𝑷𝒕𝒖𝟑) ; (47) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟒) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟐), 1 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟒] , 1 ∙ (+𝑷𝒕𝒖𝟒) ; (48) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟓) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟐), 1 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟓] , 1 ∙ (+𝑷𝒕𝒖𝟓) ; (49) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟔) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟐), 1 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟔] , 1 ∙ (+𝑷𝒕𝒖𝟔) ; (50) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟕) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟏), 1 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟕] , 1 ∙ (+𝑷𝒕𝒖𝟕) ; (51) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟖) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟐), 1 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟖] , 1 ∙ (+𝑷𝒕𝒖𝟖) ; (52) (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟗) → 1 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒅𝒔𝟐), 1 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟗] , 1 ∙ (+𝑷𝒕𝒖𝟗) ; (53) (𝒌𝑾𝒉: 𝑷𝒑) → 3 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒕𝒖𝟏), 1 ∙ [𝑭𝒕𝒖𝟏, 𝑷𝒑], 1 ∙ (+𝑷𝒑) ; (54) (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒕𝒖𝟏) → 1.05 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒅𝒔𝟐), 20 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟏), 1 ∙ [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟏], 1 ∙ (+𝑭𝒕𝒖𝟏) ; (55) (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒕𝒖𝟐) → 1.01 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒅𝒔𝟏), 20 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟐), 1 ∙ [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟐], 1 ∙ (+𝑭𝒕𝒖𝟐) ; (56) (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒕𝒖𝟑) → 1.02 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒅𝒔𝟏), 20 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟐), 1 ∙ [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟑], 1 ∙ (+𝑭𝒕𝒖𝟑) ; (57) (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒅𝒔𝟐) → 1.01 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒅𝒔𝟏), 30 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟐), 1 ∙ [𝑭𝒅𝒔𝟏, 𝑭𝒅𝒔𝟐], 1 ∙ (+𝑭𝒅𝒔𝟐) ; (58) (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒅𝒔𝟏) → 1.01 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒑𝒑), 40 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟏), 1 ∙ [𝑭𝒑𝒑, 𝑭𝒅𝒔𝟏], 1 ∙ (+𝑭𝒅𝒔𝟏) ; (59) (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒑𝒑) → 2.9 ∙ (𝑻𝒐𝒏𝑪𝒓𝒖𝒅𝒆𝑶𝒊𝒍: 𝑭𝒔), 50 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟒), 1 ∙ [𝑭𝒔, 𝑭𝒑𝒑], 1 ∙ (+𝑭𝒑𝒑). (60) As seen, the URs (43) − (44) represent knowledge about the PTDSs “1” and “2”, which are located respectively at the points 𝑷𝒕𝒅𝒔𝟏 and 𝑷𝒕𝒅𝒔𝟐, and are connected by the PTLs, represented by the multiobjects 1 ∙ [𝑷𝒑, 𝑷𝒕𝒅𝒔𝟏] and 1 ∙ [𝑷𝑷, 𝑷𝒕𝒅𝒔𝟐], with the power plant located at the place 𝑷𝒑; there are no valuable losses of the EP during it’s transmission from the PP to both PTDSs, so the same amount of the EP which is given into any PTL by the PP is received by a PTDS; hence, the multiplicities of the object (𝒌𝑾𝒉: 𝑷𝒑) in both URs (43) and (44) are equal to 1. The multiobjects 1 ∙ (+𝑷𝒕𝒅𝒔𝟏) and 1 ∙ (+𝑷𝒕𝒅𝒔𝟐) represent a fact that both PTDSs would be in active states to receive the EP from the producing it power plant and to deliver the EP to the connected with them power terminal units or PTDSs. The knowledge about PTUs “1” − “9”, connected with the respective PTDSs in full accordance with the graph representation of the considered segment of the EI, is represented by the URs (45) − (53). The UR (54) represents, that the power plant may produce one kilowatt∙hour consuming for this objective 3 tons of the fuel (represented by the multiobject 3 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒕𝒖𝟏)), receiving it via the pipe (represented by the MO 1 ∙ [𝑭𝒕𝒖𝟏, 𝑷𝒑]) from the fuel terminal unit “1” located at the point 𝑭𝒕𝒖𝟏, and being in the active state, that is represented by the MO 1 ∙ (+𝑷𝒑). The URs (55)−(57) represent knowledge about the fuel terminal units “1” − “3”. The UR (55) represents the knowledge about the resources necessary to the FTU “1” for receiving one ton of fuel from the fuel distributing station “2” located at the place 𝑭𝒅𝒔𝟐 via the pipe represented by the MO 1 ∙ [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟏]. Due to the fuel losses during transfer, the FDS “2”, delivering the fuel to the FTU “1”, gives into the pipe, represented by the MO 1 ∙ [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟏], 1.05 ton of the fuel, that is represented by the MO 1.05 ∙ (𝑻𝒐𝒏𝑭𝒖𝒆𝒍: 𝑭𝒅𝒔𝟐). The FTU “1” to receive one ton of the fuel consumes 20 kilowatt∙hours of the EP from the power terminal unit located at the point 𝑷𝒕𝒖𝟏, that is represented by the MO 20 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟐). And, as usual, the FTU “1” must be in the active state, that is represented by the MO 1 ∙ (+𝑭𝒕𝒖𝟏). The URs (56)−(57) in the same manner represent the knowledge about the fuel terminal units “2” and “3” which are provided by the EP from the PTU “2”, and this PTU consumes the same 20 kilowatt∙hours for one ton of the received fuel. The URs (58)−(59) represent the similar knowledge about the fuel distributing stations “1” and “2” provided by the EP from the PTUs “3” and “2” respectively; the FDS “1” consumes 40 kilowatt∙hours of EP from the PTU “7” located at the point 𝑷𝒕𝒖𝟑, that is represented by the MO 40 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟑), and the FDS“2” consumes 30 kilowatt∙hours of EP from the PTU “2”, located at the point 𝑷𝒕𝒖𝟐, that is represented by the MO 30 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟐). The FDS “2” receives the fuel from the FDS “1” via the pipe represented by the MO 1 ∙ [𝑭𝒅𝒔𝟏, 𝑭𝒅𝒔𝟐]. The FDS “1”, in turn, receives the fuel from the fuel producing plant via the pipe represented by the MO 1 ∙ [𝑭𝒑𝒑, 𝑭𝒅𝒔𝟏] consuming 40 kilowatt∙hours of EP from the PTU “3”, located at the point 𝑷𝒕𝒖𝟑, and this is represented by the MO 40 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟑). At last, the UR (60) represents the knowledge about the FPP which is capable to produce one ton of the fuel receiving 2.9 tons of crude oil from the fuel storage via a pipe represented by the MO 1 ∙ [𝑭𝒔, 𝑭𝒑𝒑] and consuming 50 kilowatt∙hours of EP from the PTU “4”, located at the point 𝑷𝒕𝒖𝟒, and this is represented by the MO 50 ∙ (𝒌𝑾𝒉: 𝑷𝒕𝒖𝟒). Finally, as seen, the considered segment of the EI, consuming crude oil from the fuel storage, provides external consumers by the electric power and the fuel, respectively, via the PTUs “5”, “6” and “7”, and via the FTUs “2” and “3”. ∎ A resource base of any EI may be represented as a multiset 𝑣𝐸 including multiobjects of the following three types: 1) 𝑚 ∙ (𝑟𝑒𝑠 : 𝑝) for all fuel storages entering a considered EI, that means 𝑚 units of materiel resource (PEC or produced fuel) 𝑟𝑒𝑠 are available at some FS located at a place 𝑝; 2) 𝑁 ∙ [𝑝, 𝑝′] for all links having place in a considered EI, that means a value 𝑁 is a bandwidth (throughput capacity) of a link [𝑝, 𝑝′], i.e. a maximal amount of EP or materiel resource, which may be transmitted (transferred) via this link during a considered time period (in the case [𝑝, 𝑝′] is a PTL this amount is measured in kilowatt∙hours whilst in the case [𝑝, 𝑝′] is a pipe this amount may be measured in barrels, cubic meters, kilograms, tons etc.); 3) 𝐿 ∙ (+𝑥) for all elements of a considered EI, thus establishing for any such element a maximal number of operation cycles which might be executed by it at a considered time period (in other words, 𝐿 is fixing a maximal productivity of an element 𝑥; a multiobject 𝐿 ∙ (+𝑥) will be referred below as an operation resource of an element 𝑥). So in fact a resource base of any EI includes not only materiel resources (primary and produced energy carriers), but also operation resources of it’s elements, as well as throughput capacities of it’s links. Example 5. The resource base of the segment of the EI considered in the previous Example 4 and corresponding to the knowledge represented by the Table 1, is as follows: 𝑣𝐸 = {100 ∙ (𝑻𝒐𝒏𝑪𝒓𝒖𝒅𝒆𝑶𝒊𝒍: 𝑭𝒔), 1000 ∙ [𝑷𝒑, 𝑷𝒕𝒅𝒔𝟏], 1100 ∙ [𝑷𝒑, 𝑷𝒕𝒅𝒔𝟐], 200 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟏], 300 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟐], 400 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟑], 100 ∙ [𝑷𝒕𝒅𝒔𝟏, 𝑷𝒕𝒖𝟕], 200 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟒], 300 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟓] , 300 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟔], 200 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟖], 100 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟗], 200 ∙ [𝑭𝒔, 𝑭𝒑𝒑], 200 ∙ [𝑭𝒑𝒑, 𝑭𝒅𝒔𝟏], 100 ∙ [𝑭𝒅𝒔𝟏, 𝑭𝒅𝒔𝟐], 50 ∙ [𝑭𝒅𝒔𝟏, 𝑭𝒕𝒖𝟐], 50 ∙ [𝑭𝒅𝒔𝟏, 𝑭𝒕𝒖𝟑], (61) 100 ∙ [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟏], 100 ∙ [𝑭𝒕𝒖𝟏, 𝑷𝒑], 100 ∙ (+𝑷𝒑), 100 ∙ (+𝑷𝒕𝒅𝒔𝟏), 100 ∙ (+𝑷𝒕𝒅𝒔𝟐), 100 ∙ (+𝑷𝒕𝒖𝟏), 100 ∙ (+𝑷𝒕𝒖𝟐), 100 ∙ (+𝑷𝒕𝒖𝟑), 100 ∙ (+𝑷𝒕𝒖𝟒), 100 ∙ (+𝑷𝒕𝒖𝟓), 100 ∙ (+𝑷𝒕𝒖𝟔), 100 ∙ (+𝑷𝒕𝒖𝟕), 10 ∙ (+𝑭𝒕𝒖𝟏), 10 ∙ (+𝑭𝒕𝒖𝟐), 10 ∙ (+𝑭𝒕𝒖𝟑), 10 ∙ (+𝑭𝒅𝒔𝟐), 10 ∙ (+𝑭𝒅𝒔𝟏), 10 ∙ (+𝑭𝒑𝒑)}. As seen, the fuel storage entering the considered segment of an EI contains 100 tons of crude oil; the PTL connecting the power plant and the PTDS “1” during a considered time period provides transmission no more than 1000 kilowatt∙hours of EP that is represented by the MO 1000 ∙ [𝑷𝒑, 𝑷𝒕𝒅𝒔𝟏]; the PTL connecting the PP and the PTDS “2” provides transmission no more than 1100 kilowatt∙hours of EP that is represented by the MO 1100 ∙ [𝑷𝒑, 𝑷𝒕𝒅𝒔𝟐]; similarly are represented the upper threshold values of bandwidths of all other PTLs of the considered segment of the EI. The pipe, connecting the fuel storage and the fuel producing plant, provides delivery of no more than 200 tons of crude oil that is represented by the MO 200 ∙ [𝑭𝒔, 𝑭𝒑𝒑]; the pipe, connecting the fuel producing plant and the fuel distributing station “1”, provides delivery of no more than 200 tons of the fuel that is represented by the MO 200 ∙ [𝑭𝒑𝒑, 𝑭𝒅𝒔𝟏]; similarly are represented the upper threshold values of throughput capacities of all other pipes of the considered segment of the EI. Any element of the ElcI, entering this segment, during a considered period of time may execute 100 operation cycles, that is represented by the MOs 100 ∙ (+𝑷𝒑), … , 100 ∙ (+𝑷𝒕𝒖𝟗); any element of the FI, entering this segment, during a considered period of time may execute 10 operation cycles, that is represented by the multiobjects 10 ∙ (+𝑭𝒕𝒖𝟏), … , 10 ∙ (+𝑭𝒑𝒑). ∎ After specifying a resource base, an EI 𝐸 may be considered as a free industrial system 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > in the sense [7]. Similarly, a demand on electric power and fuel (an order to be completed in the sense of [7]) may be represented as a multiset 𝑞𝐸 containing multiobjects like 𝑛 ∙ (𝒌𝑾𝒉: 𝑝), representing 𝑛 kilowatt∙hours which would be delivered to a consumer located at a place 𝑝 where some PTU providing this delivery is located, and multiobjects like 𝑚 ∙ (𝑟𝑒𝑠: 𝑝), representing 𝑚 units of a fuel (or any other materiel resource) 𝑟𝑒𝑠 which would be delivered to a consumer located at a place 𝑝 where an FTU providing this delivery is located. As a result, an EI providing delivery of needed to consumers amounts of power and fuel may be considered as an industrial system 𝐸𝑞 =< 𝑞, 𝑅𝐸 , 𝑣𝐸 > assigned to an order 𝑞 in the sense [7]. Following [7], this representation of an IS implies a filtering unitary multiset grammar 𝑆𝑞 =< 𝑞, 𝑅𝐸 , 𝐹𝐸 >, where 𝐹𝐸 = { 𝑎 ≤ 𝑛 | 𝑛 ∙ 𝑎 ∈ 𝑣𝐸 } ∪ {𝑎 = 0 | 𝑎 ∈ 𝐴𝑆̅ & 𝑎 ⋶ 𝑣𝐸 }, (62) in such a way that this FUMG generates a set of terminal multisets each representing some collection of resources sufficient for an order 𝑞 completion by some definite cooperation of manufacturing devices (the second operand of a join is obligatory to eliminate ways of an order completion which satisfy restrictions implied by an available resource base of an EI, but need some additional resources which are absent at an RB at all). As now may be seen, a unitary multiset grammar 𝑆𝑞 =< 𝑞, 𝑅𝐸 > defines a set 𝑉̅𝑆𝑞 of terminal multisets each having a form {𝑀1 ∙ (𝑟𝑒𝑠𝑖1 : 𝑝𝑖1 ), … , 𝑀𝑠 ∙ (𝑟𝑒𝑠𝑖𝑡 : 𝑝𝑖𝑡 ), 𝑁1 ∙ [𝑝𝑗1 , 𝑝′ 𝑗 ],…, 𝑁𝑢 ∙ [𝑝𝑗𝑢 , 𝑝′𝑗𝑢 ], 1 (63) 𝐿1 ∙ (+𝑥𝑘1 ), … , 𝐿𝑧 ∙ (+𝑥𝑘𝑧 )}, where 𝑀1 ,…, 𝑀𝑠 are amounts of, respectively, resources 𝑟𝑒𝑠𝑖1 ,…, 𝑟𝑒𝑠𝑖𝑡 (PECs stored at FSs, fuels, produced by FPPs, as well as EP, produced by PPs) to be available at places 𝑝𝑖1 , … , 𝑝𝑖𝑡 (via PTUs, FTUs, or directly from fuel storages); 𝑁1 … , 𝑁𝑢 are amounts of energy carriers and electric power to be transferred (transmitted) via, respectively, links [𝑝𝑗1 , 𝑝′𝑗1 ],…,[𝑝𝑗𝑢 , 𝑝′𝑗𝑢 ] (PTLs and pipes) during a considered time period; 𝐿1 … , 𝐿𝑧 are numbers of operation cycles of, respectively, facilities 𝑥𝑘1 , … , 𝑥𝑘𝑧 involved in a completion of an order 𝑞. So every TMS 𝑣 ∈ 𝑉̅𝑆𝑞 corresponds to some specific way of an order 𝑞 completion (in a general case there may be several ways identical by resource consumption and facilities involvement). We shall represent an EI current resource base 𝑣𝐸 as a sum of three multisets 𝑝 𝑣𝐸 = 𝑣𝐸𝑟𝑒𝑠 +𝑣𝐸 +𝑣𝐸𝑥 , (64) the first 𝑣𝐸𝑟𝑒𝑠 = { 𝑴1 ∙ (𝑟𝑒𝑠𝒊𝟏 : 𝑝𝒊𝟏 ), … , 𝑴𝒔 ∙ (𝑟𝑒𝑠𝒊𝒕 : 𝑝𝒊𝒕 ) } (65) representing amounts of resources having place at an EI fuel storages, the second 𝑝 𝑣𝐸 = { 𝑵1 ∙ [𝑝𝒋𝟏 , 𝑝′ 𝒋 ],…, 𝑵𝒖 ∙ [𝑝𝒋𝒖 , 𝑝′ 𝒋 ]} (66) 𝟏 𝒖 representing current bandwidths and throughput capabilities of an EI links, and the third 𝑣𝐸𝑥 = { 𝑳𝟏 ∙ (+𝑥𝒌𝟏 ), … , 𝑳𝒛 ∙ (+𝑥𝒌𝒛 )} (67) representing current operation resource of an EI facilities. (Bold indices 𝒊𝟏 , . . , 𝒊𝒕 , 𝒋𝟏 , … , 𝒋𝒖 , 𝒌𝟏 , … , 𝒌𝒛 , used in (64) −(67), differ from ordinary indices 𝑖1 , . . , 𝑖𝑡 , 𝑗1 , … , 𝑗𝑢 , 𝑘1 , … , 𝑘𝑧 , used in (63)). 5. Cyclicity of FUMGs, representing energy infrastructures, and their finitarization Let us note, that industrial systems are represented through a technological interpretation of unitary rules [3, 4, 7], or in other words, through their capability to manufacture (assemble) some complex objects from their components until some atomary (non-splitted) elements (spare parts, microchips, etc.); thus FUMGs representing ISs are essentially acyclic, and, hence, STMSs generated by their application, are finite. Unlike industrial systems, energy infrastructures operate in such a way that it’s fuel segment (namely, FI) consumes EP generated by it’s electricity segment (namely, ElcI), whilst the last one consumes fuel necessary for EP production. Thus FUMGs representing EIs are essentially cyclic, and sets of multisets generated by their application are in a general case infinite: for a core UMG 𝑆′𝑞 =< 𝑞, 𝑅𝐸 > of a FUMG 𝑆𝑞 =< 𝑞, 𝑅𝐸 , 𝐹𝐸 > it would be valid 𝑉𝑆′𝑞 = {∅} (68) and, simultaneously, |𝑉𝑆′𝑞 | = ∞. (69) hence direct application of an ISs multigrammatical representation and criterial base to EIs is in fact impossible. So a task is to find such local correction of the aforementioned representation of ISs which provide finitarity of FUMGs representing EIs. Such correction will be called finitarization of FUMGs. Here we propose a simple solution of this problem based on the so called terminalization of non-terminal objects introduced in [7] as a tool of modelling ISs, which resource bases contain not only primary (non-splitted) components of objects specified by an order to an IS, but also components, manufactured by an IS beginning from the aforementioned primary ones at previous steps of it’s operation. Namely, we shall extend a set of URs 𝑅𝐸 in a following way. Let 𝑅𝐸 contains an unitary rule (𝒌𝑾𝒉: 𝑥) → 𝑋, (70) where 𝑋 is a non-empty body. We shall join to 𝑅𝐸 an unitary rule (𝒌𝑾𝒉: 𝑥) → 𝑘 ∙ (𝑟𝑒𝑠: 𝑥 ′ ), (71) that means one kilowatt∙hour would appear at a location 𝑥 as a result of consumption 𝑘 units of resource 𝑟𝑒𝑠 located at a place 𝑥′, and, that is most essential, (𝑟𝑒𝑠: 𝑥′) is a terminal object, that means 𝑅𝐸 does contain no one UR with a header (𝑟𝑒𝑠: 𝑥′); the last, in turn, means, that there is an alternative way of such appearance, not involving chain of mutual demands determined by a body 𝑋 of UR (70). In most cases such resource 𝑟𝑒𝑠 is power, accumulated at previous steps of operation of an EI or generated by some initiating action or operation (for example, activation of a car ignition system). In the first case power storages (PSs) similar to fuel storages are presumed, and, like FSs, they may be represented by multiobjects 𝑛 ∙ (𝒌𝑾𝒉: 𝑥) ∈ 𝑣𝐸 (72) that means a PS located at a place 𝑥 may provide on demand up to 𝑛 kilowatt∙hours. The second case (power generation by some initiating action) is simply reduced to the first one by including to a resource base the same multiobject as in (72), that reflects an obstacle, that a source of the aforementioned action is, finally, is also some kind of a power storage. Thus, introducing by (71)−(72) a concept of a power storage, which, in fact, fully reflects essence of real processes of power supply, we have proposed the simplest way of finitarization of FUMGs representing EIs. Now, evidently, despite a set 𝑉𝑆′𝑞 remains infinite, a set 𝑉𝑆′𝑞 in a general case would be non-empty, thus representing at least one way of an order 𝑞 completion by application of a priori accumulated power; from the mathematical point of view, this means that a core UMG 𝑆′𝑞 =< 𝑞, 𝑅𝐸 > of a FUMG 𝑆𝑞 =< 𝑞, 𝑅𝐸 , 𝐹𝐸 > generates at least one terminal TMS. Now we are ready to consider the main result of this paper being a criterial base for the assessment of vulnerability of energy infrastructures to destructive impacts. 6. Criteria of vulnerability of an energy infrastructure to a destructive impact Let us begin from the initial task, which verbal formulation is as follows: given amounts of primary energy carriers at fuel storages of an EI and demand on an electric power and fuels from it’s external consumers (an order to be completed by an EI), to assess whether an EI is or is not capable to complete an order (i.e. to provide these consumers by required amounts of EP and fuels). Due to the introduced techniques of EIs representation, now to solve this task it is sufficient to apply the criterion [7], proposed regarding industrial systems, to energy infrastructures. Statement 1. An energy infrastructure 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > is not capable to complete an assigned order 𝑞, if (∀𝑣 ∈ 𝑉𝑆𝑞 ) 𝑣𝐸 ⊂ 𝑣, (73) where 𝑆𝑞 =< 𝑞, 𝑅𝐸 >.∎ Speaking informally, an EI 𝐸 is not capable to complete an assigned order 𝑞, if there exists no one way of generation (production) and delivery of necessary amounts of EP and fuels, consuming for this objective such amounts of primary energy carriers which are not greater than available at fuel storages of this EI, and also capabilities of EI facilities and links are sufficient for these generation (production) and delivery. Following [7], this criterion may be represented by applying a respective filtering unitary multiset grammar 𝑆𝑞 =< 𝑞, 𝑅𝐸 , 𝐹𝐸 >, where 𝐹𝐸 = { 𝑎 ≤ 𝑛 | 𝑛 ∙ 𝑎 ∈ 𝑣𝐸 } ∪ {𝑎 = 0 | 𝑎 ∈ 𝐴𝑆̅ & 𝑎 ⋶ 𝑣𝐸 } (74) (the second operand of a join is obligatory to eliminate ways of order completion which satisfy restrictions implied by an available resource base of an EI, but need some additional resources which are absent at an RB at all). Statement 2. Energy infrastructure 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > is not capable to complete an assigned order 𝑞, if 𝑉𝑆𝑞 = {∅}, (75) where 𝑆𝑞 =< 𝑞, 𝑅𝐸 , 𝐹𝑞 >.∎ All the said forms a background for the strict consideration of a task of an assessment of vulnerability of EIs to destructive impacts. Following [7], we shall represent an impact as a multiset 𝛥𝑣 which determines eliminated by this impact capabilities of an EI elements (facilities and links) and stored at FSs amounts of primary energy carriers. After such impact application a resource base 𝑣𝐸 of an EI becomes 𝑣𝐸 - 𝛥𝑣. Such representation in a general case provides any possible variants of impact, which may destroy EI elements, reduce bandwidths (throughput capacities) of links PTLs and pipes, as well as reduce amounts of PECs in FSs. Namely, 𝑚 ∙ (𝑟𝑒𝑠 : 𝑝) ∈ 𝛥𝑣 (76) means, that an impact eliminates 𝑚 units of a resource from an FS located at a place 𝑝. Similarly, 𝑛 ∙ [𝑝, 𝑝′] ∈ 𝛥𝑣 (77) means, that an impact reduces for 𝑚 units a maximal amount of EP or fuel which may be transmitted (transferred) at a considered time period via a PTL (pipe) with start point 𝑝 and final point 𝑝′. Finally, to represent destruction of any producing or transmitting (transferring) facility (PP, FTP, PTDS, FTDS, PTU, FTU) we may apply the same techniques including to an MS 𝛥𝑣 multiobjects like 𝑙 ∙ (+𝑥) representing that an element of EI would be affected, and a result of this action would be reduction of an operation resource of this element by 𝑙 units. Obviously, a case of an entire destruction of any component of an EI may be easily represented by inclusion to a multiset 𝛥𝑣 an object 𝑵 ∙ (𝑎), where 𝑵 is a number maximal for the used implementation of FUMGs algorithmics, so for any 𝑘 {𝑘 ∙ (𝑎)} - {𝑵 ∙ (𝑎)} = {∅} (78) is valid. Example 6. Let the destructive impact destroys facilities PTDS ”1” and PTU “7” of the considered in the previous Example 5 segment of an EI as well as reduces amount of the crude oil at the fuel storage by 20 tons, and also reduces bandwidths (throughput capacities) of: PTL [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟒] by 100 kilowatt∙hours, PTL [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟓] by 200 kilowatt∙hours, and pipe [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟏] by 10 tons of fuel. So the result of this impact will be 𝑣𝐸 - 𝛥𝑣 = 𝑣𝐸 - {10000 ∙ (+𝑷𝒕𝒅𝒔𝟏), 10000 ∙ (+𝑷𝒕𝒖𝟕), 20∙ (𝑻𝒐𝒏𝑪𝒓𝒖𝒅𝒆𝑶𝒊𝒍: 𝑭𝒔), (79) 100 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟒], 200 ∙ [𝑷𝒕𝒅𝒔𝟐, 𝑷𝒕𝒖𝟓], 10 ∙ [𝑭𝒅𝒔𝟐, 𝑭𝒕𝒖𝟏]}, where 𝑵 = 10000. ∎ Let us begin from the simplest case when an impact is applied to an EI before a beginning of an order 𝒒 completion. Evidently, an impact 𝛥𝑣 transforms an EI 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > to an affected EI 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 - 𝛥𝑣 >. Now to obtain a necessary criteria of vulnerability of energy infrastructures to destructive impacts, it is sufficient to apply Statements 1 and 2 to an affected EI. Statement 3. An energy infrastructure 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > is vulnerable to an impact 𝛥𝑣, applied before a beginning of an assigned order 𝑞 completion, if (∀𝑣 ∈ 𝑉𝑆𝑞 ) 𝑣𝐸 - 𝛥𝑣 ⊂ 𝑣 (80) where 𝑆𝑞 =< 𝑞, 𝑅𝐸 >.∎ Verbally, if no one way of an order 𝑞 completion is implementable (any way needs additional resources regarding available after an impact), then an energy infrastructure is vulnerable to an applied impact. Similarly to the Statement 3, this criterion may be represented by the application of FUMGs. Statement 4. An energy infrastructure 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > is vulnerable to an impact 𝛥𝑣, applied before a beginning of an assigned order 𝑞 completion, if 𝑉𝑆𝑞 = {∅}, (81) where 𝑆𝑞 =< 𝑞, 𝑅𝐸 , 𝐹𝐸′ >, and 𝐹𝐸′ = { 𝑎 ≤ 𝑛 | 𝑛 ∙ 𝑎 ∈ 𝑣𝐸 - 𝛥𝑣} ∪ {𝑎 = 0 | 𝑎 ∈ 𝐴𝑆̅ & 𝑎 ⋶ 𝑣𝐸 - 𝛥𝑣}. ∎ (82) Let us consider now a more general case, when an impact is applied to an EI at some time moment inside time period of an order completion. Just to this moment some part 𝛥𝑞 of an order 𝑞 may be completed, as well as a respective part 𝛥𝑣𝐸 of an EI resource base would be already consumed. If so, then there is not difficult to formulate statements being a corollaries of the Statements 3 and 4 and representing criteria of vulnerability of an EI affected by a destructive impact inside a time period of an order completion. Statement 5. An energy infrastructure 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > is vulnerable to an impact 𝛥𝑣, applied inside a time period of a completion an assigned order 𝑞, when a part 𝛥𝑞 of this order is already completed and a part 𝛥𝑣𝐸 of an EI resource base is already consumed, if (∀𝑣 ∈ 𝑉𝑆𝑞-𝛥𝑞 ) 𝑣𝐸 - 𝛥𝑣𝐸 - 𝛥𝑣 ⊂ 𝑣. ∎ (83) Statement 6. An energy infrastructure 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 > is vulnerable to an impact 𝛥𝑣, applied inside a time period of a completion an assigned order 𝑞, when a part 𝛥𝑞 of this order is already completed and a part 𝛥𝑣𝐸 of an EI resource base is already consumed, if 𝑉𝑆𝑞-𝛥𝑞 = {∅}, (84) where 𝑆𝑞-𝛥𝑞 =< 𝑞, 𝑅𝐸 , 𝐹𝐸′′ >, and 𝐹𝐸′′ = { 𝑎 ≤ 𝑛 | 𝑛 ∙ 𝑎 ∈ 𝑣𝐸 - 𝛥𝑣𝐸 - 𝛥𝑣} ∪ {𝑎 = 0 | 𝑎 ∈ 𝐴𝑆̅ & 𝑎 ⋶ 𝑣𝐸 - 𝛥𝑣𝐸 - 𝛥𝑣}. ∎ (85) Now we may consider a more complicated case of EIs which topology is designed and resource base is maintained in such a way that if some destructive impact is applied before or during order completion, and this impact makes an EI not capable to complete this order, then an EI recovers itself by activation some prepared in advance amounts of operation resources as well as by application some prestored amounts of materiel resources. 7. Modelling reservation and recovery of energy infrastructures Taking into account a possibility of application of destructive impacts of various nature to components of energy infrastructure, an EIs’ management is usually preparing in advance some additional reserved facilities and primary or produced resources, which are made available promptly after an impact is detected, and such measure in many cases provides as effective as possible mitigation of consequences of the aforementioned impact (up to making this order feasible by the adjusted itself EI). A background of such adaptability is some redundancy implanted to an EI before or during it’s operation [30-34]. The most usual measure implemented by EIs’ designers and management are the so called backup power systems, providing EP generation for a time periods when the affected segments of EIs are recovered, and also bypasses, providing electricity or fuel flows by some workarounds if a preordered routes are broken by an impact. It is not so difficult to apply the UMGs to represent the described opportunity. Namely, it is sufficient to join to an initial set of URs, representing a topology of an electricity infrastructure, unitary rules reflecting an alternative ways of EP transmission. Namely, for any UR (18) representing a PTU, located at a point 𝑝𝑡𝑢, it is sufficient to join to a set 𝑅𝐸 one more UR (𝒌𝑾𝒉: 𝑝𝑡𝑢) → 𝑛′ ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠 ′ ), 𝑛′ ∙ [𝑝𝑡𝑑𝑠′, 𝑝𝑡𝑢] , (86) representing a fact that this PTU may receive EP not only from a PTDS located at a point 𝑝𝑡𝑑𝑠, but also from a PTDS located at a point 𝑝𝑡𝑑𝑠′. (It is assumed that there is a technological solution providing such opportunity). In a general case there may be 𝑚 ≥ 1 such alternative PTDSs capable to deliver EP to this PTU, and this is possible regarding any PTU entering a considered EI. Similarly, the same technique may be applied to PTDSs. To any UR entering a set (20) and representing a PTDS, located at a point 𝑝𝑡𝑑𝑠𝑖 , it is sufficient to join to a set 𝑅𝐸 one more UR (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠𝑖 ) → 𝑛′𝑖 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠 ′ ), 𝑛′ 𝑖 ∙ [𝑝𝑡𝑑𝑠′, 𝑝𝑡𝑑𝑠𝑖 ], (87) representing a fact that this PTDS may receive EP not only from a PTDS located at a point 𝑝𝑡𝑑𝑠, but also from a PTDS located at a point 𝑝𝑡𝑑𝑠′. As in the case of PTUs, there may be 𝑚 ≥ 1 such alternative PTDSs capable to deliver EP to this PTDS, and this is possible regarding any PTDS entering a considered EI. The described technique without any changes may be applied also to PTUs and PTDSs, connected directly with additional (reserve) power plants: (𝒌𝑾𝒉: 𝑝𝑡𝑑𝑠𝑗 ) → 𝑛𝑗 ∙ (𝒌𝑾𝒉: 𝑝𝑝′ ), 𝑛𝑗 ∙ [𝑝𝑝′, 𝑝𝑡𝑑𝑠𝑗 ], (88) (𝒌𝑾𝒉: 𝑝𝑡𝑢𝑘 ) → 𝑛𝑘 ∙ (𝒌𝑾𝒉: 𝑝𝑝′′), 𝑛𝑘 ∙ [𝑝𝑝′′, 𝑝𝑡𝑢𝑘 ]. (89) These URs represent the facts, that a PTDS, located at a point 𝑝𝑡𝑑𝑠𝑗 , may receive EP not only from a PP located at a point 𝑝𝑝, as defined by a respective UR from a set (21), but also from a PP located at a place 𝑝𝑝′, as well as PTU, located at a point 𝑝𝑡𝑢𝑘 and entering a set of URs (22), may receive EP from some PP located at a place 𝑝𝑝′′, which may differ from 𝑝𝑝′ or be the same. As in all considered above cases, there may be 𝑚 ≥ 1 such alternative power plants capable to deliver EP to these PTUs and PTDSs, and this is possible regarding any PTU and PTDS connected with several power plants. Any reserve power plant, entering a considered ElcI and located at a point 𝑝𝑝′, may be represented by an UR (𝒌𝑾𝒉: 𝑝𝑝′) → 𝑛′1 ∙ (𝑟𝑒𝑠′1 : 𝑝′1 ), … , 𝑛′𝑘′ ∙ (𝑟𝑒𝑠′𝑘′ : 𝑝′𝑘′ ) , (90) where, as in (23), 𝑛′1 , … , 𝑛′𝑘′ are amounts of resources 𝑟𝑒𝑠′1 , … , 𝑟𝑒𝑠′𝑘′ , which must be delivered to locations 𝑝′1 , … , 𝑝′𝑘′ respectively in order to generate one kilowatt∙hour of electrical power at a location 𝑝𝑝′ , from which, in turn, it may be delivered by PTLs to PTDSs (PTUs), closest to this PP. Let us note, that a reservation may be implemented not only by inclusion to an ElcI some additional power plants but also by implementation of alternative ways of EP generation and associated with them resources, by which a PP must be supplied. In this case to an UR (23) a unitary rule (𝒌𝑾𝒉: 𝑝𝑝) → 𝑛′1 ∙ (𝑟𝑒𝑠′1 : 𝑝′1 ), … , 𝑛′𝑘′ ∙ (𝑟𝑒𝑠′𝑘′ : 𝑝′𝑘′ ) , (91) with the same header (𝒌𝑾𝒉: 𝑝𝑝) and alternative body, representing a respective supply set, necessary for this way implementation, is joined to a set 𝑅𝐸 . As may be seen, due to an application of alternating UMGs it is quite easy to represent electric grids of any complexity, not only of tree-like structure, as it was considered above in the Section 4. Similarly may be represented reservation of a fuel infrastructure. Namely, for any UR (24) representing a FTU, located at a point 𝑓𝑡𝑢, it is sufficient to join to a set 𝑅𝐸 one more UR (𝑟𝑒𝑠: 𝑓𝑡𝑢) → 𝑛′ ∙ (𝑟𝑒𝑠: 𝑓𝑑𝑠 ′ ), 𝑚′ ∙ (𝒌𝑾𝒉: 𝑓𝑡𝑢), 𝑛 ∙ [𝑓𝑑𝑠′, 𝑓𝑡𝑢], (92) representing a fact that this FTU may receive a resource 𝑟𝑒𝑠 not only from a FDS located at a point 𝑓𝑑𝑠, but also from a FDS located at a point 𝑓𝑑𝑠′. (As in the case of ElcI, it is assumed that there is a technological solution providing such opportunity). In a general case there may be 𝑚 ≥ 1 such alternative FDSs capable to deliver a resource 𝑟𝑒𝑠 to this FTU, and this is possible regarding any FTU entering a considered FI. Reservation of distributing facilities (namely, FDSs) of fuel infrastructure may be represented similarly to reservation of TDSs: (𝑟𝑒𝑠: 𝑓𝑑𝑠𝑖 ) → 𝑛′𝑖 ∙ (𝑟𝑒𝑠: 𝑓𝑑𝑠′), 𝑚′1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢𝑖 ), 𝑛′1 ∙ [𝑓𝑑𝑠′, 𝑓𝑑𝑠𝑖 ]. (93) Any such UR represents the fact, that an FDS, located at a point 𝑓𝑑𝑠𝑖 , may receive required amounts of resource 𝑟𝑒𝑠 not only from an FDS located at a point 𝑓𝑑𝑠, as defined by a respective UR from a set (25), but also from an FDS located at a place 𝑓𝑑𝑠′. As in all considered above cases, there may be 𝑚 ≥ 1 such alternative FDSs capable to deliver resource 𝑟𝑒𝑠 to this FDS, and such opportunity is possible regarding any FDS connected with several supplying it FDSs. At last, any reserve FPP, producing fuel 𝑟𝑒𝑠 used by power plants for EP generation, located at a point 𝑓𝑝𝑝′, may be represented by an UR (𝑟𝑒𝑠: 𝑓𝑝𝑝′) → 𝑛′ ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′), 𝑚′1 ∙ (𝑟𝑒𝑠′1 : 𝑓𝑠′1 ), 𝑛′1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′1 ), (94) … , 𝑚′𝑡 ∙ (𝑟𝑒𝑠′𝑡 : 𝑓𝑠′𝑡′ ), 𝑛′𝑡′ ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′𝑡′ ), where 𝑓𝑠′1 , … , 𝑓𝑠′𝑡′ are points, where fuel storages with PECs 𝑟𝑒𝑠′1 , … , 𝑟𝑒𝑠′𝑡′ are located, so namely regarding these places power terminal units would be installed, thus providing relocation of amounts of these PECs necessary to an FPP for production of one unit of fuel 𝑟𝑒𝑠 at a location 𝑓𝑝𝑝′. The aforementioned relocation would be possible if needed amounts of electric power, i.e. 𝑛′1 ,…, 𝑛′𝑡′ kilowatt∙hours, would be available at points 𝑝𝑡𝑢′1 , … , 𝑝𝑡𝑢′𝑡′ where respective PTUs are operating. In turn, to produce one unit of a fuel 𝑟𝑒𝑠 a reserve FPP itself would consume 𝑛′ kilowatt∙hours from a power terminal unit located at a point 𝑝𝑡𝑢′. Similarly to power plants, any fuel producing plant may be reserved not only by inclusion to an FI some additional FPPs but also by implementation of alternative ways of fuel producing and associated with them PECs, by which an FPP must be supplied. In this case for any UR (27) a unitary rule (𝑟𝑒𝑠: 𝑓𝑝𝑝) → 𝑛′ ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′), 𝑚′1 ∙ (𝑟𝑒𝑠′1 : 𝑓𝑠′1 ), 𝑛′1 ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′1 ), (95) … , 𝑚′𝑡 ∙ (𝑟𝑒𝑠′𝑡 : 𝑓𝑠′𝑡′ ), 𝑛′𝑡′ ∙ (𝒌𝑾𝒉: 𝑝𝑡𝑢′𝑡′ ), with the same header (𝑟𝑒𝑠: 𝑓𝑝𝑝) and alternative body, representing a respective alternative supply set, necessary for this way implementation, is joined to a set 𝑅𝐸 . As may be seen, this generalization makes possible application of the criteria (80)−(85) to a general case of EIs without any corrections. The main difference is that UMGs representing such EIs are alternating, and thus 𝑉𝑆𝑞 is a multi-element set of terminal multisets, i.e. |𝑉𝑆𝑞 | ≥ 1. However, in practice an EI operates by some subset of it’s components, and this subset as a whole has an ordinary concentric tree-like structure whilst the rest components stay in a reserve until an impact, after which some or even all of reserve components may be joined (switched) to an affected EI. To implement this approach it is sufficient to represent a reserved EI as a ternary tuple (for short “quadraple”) 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 , 𝒗𝑬 >, where 𝒗𝑬 is a reserve resource base, any part (submultiset) of which 𝜟𝒗𝑬 𝒗𝑬 may be added to an RB reduced by an impact, transforming it from 𝑣𝐸 - 𝛥𝑣 to 𝑣𝐸 - 𝛥𝑣+𝜟𝒗𝑬 . Following (64)−(67), a multiset 𝜟𝒗𝑬 may be represented 𝑝 as a sum of three non-intersecting multisets similar to MSs 𝑣𝐸𝑟𝑒𝑠 , 𝑣𝐸 , 𝑣𝐸𝑥 : 𝑝 𝜟𝒗𝑬 = 𝜟𝒗𝑟𝑒𝑠 𝑥 𝐸 +𝜟𝒗𝐸 +𝜟𝒗𝐸 , (96) the first summand 𝜟𝒗𝑟𝑒𝑠 𝐸 = { 𝜟𝑴1 ∙ (𝑟𝑒𝑠𝒊𝟏 : 𝑝𝒊𝟏 ), … , 𝜟𝑴𝒔 ∙ (𝑟𝑒𝑠𝒊𝒔 : 𝑝𝒊𝒔 ) } (97) representing amounts of resources (PECs and fuels) having place at EI reserve fuel storages, as well as amounts of EP accumulated by backup power systems (in this case 𝑟𝑒𝑠𝒊𝒋 is nothing but 𝒌𝑾𝒉 ); the second 𝑝 𝜟𝒗𝐸 = { 𝜟𝑵1 ∙ [𝑝𝒋𝟏 , 𝑝′ 𝒋 ],…, 𝜟𝑵𝒖 ∙ [𝑝𝒋𝒖 , 𝑝′ 𝒋 ]} (98) 𝟏 𝒖 representing reserve throughput capabilities of an EI (PTLs and pipes kept out of operation until an impact), and the third 𝜟𝒗𝐸𝑥 = { 𝜟𝑳𝟏 ∙ (+𝑥𝒌𝟏 ), … , 𝜟𝑳𝒛 ∙ (+𝑥𝒌𝒛 )} (99) representing reserve operation resource of an EI (facilities kept ready to operate if necessary). Thus addition to the reduced RB an MS 𝜟𝒗𝑟𝑒𝑠 𝐸 provides join to an affected EI some amounts of fuel located at reserve FSs, an MS 𝑝 𝜟𝒗𝐸 − switching to an EI transporting network some reserve links (PTLs and pipes), and, at last, an MS 𝜟𝒗𝐸𝑥 − join to an EI some reserve producing, generating and transmitting/transferring facilities. After this addition criteria (80) − (82) may be applied to an EI 𝐸 =< {∅}, 𝑅𝐸 , 𝑣𝐸 - 𝛥𝑣+𝜟𝒗𝑬 > and an order 𝑞. If this EI is not vulnerable then there exists an opportunity to recover it after impact, and thus there arises naturally a task of computation “the best” of all possible multisets 𝜟𝒗𝑬 𝒗𝑬 providing recovery of the affected EI to a state sufficient for an order completion. Otherwise it is clear that available reserve is insufficient for EI recovery and order completion. 8. Modelling rechargeable power storages and their application Until now, introducing the concept of a power storage, we have not determined nor verbally, nor strictly, how power is accumulated in any specific PS, as well as how the last is recharged after or during order completion (for example, in a case of a car accumulator it is recharged during or, more correct, by a car motion). Let us consider a case of rechargeable power storages and their multigrammatical modelling.. Namely, we shall associate with any order 𝑞, which objective is meeting a demand on predefined by an external consumer collections of resources located at predefined places, a so called internal order 𝑞′, which objective is addition of a collection 𝑞′ to a resource base, i.e. full or partial (or even redundant) replenishment of resources spent whilst order 𝑞 completion. The simplest way of an internal order interpretation is a replacement of an RB 𝑣𝐸 , remained after external order 𝑞 completion, by 𝑣𝐸 +𝑞′. However, such approach is not satisfactory, because it, in fact, applies a concept of an energy infrastructure as an open system – a collection 𝑞′ is not a part of EI resource base and is applied from systems which are external regarding EI. The second reason for rejection of this approach is that a collection 𝑞′ is not at all correlated with an order 𝑞; it would be assigned to by an EI control system in some arbitrary way (the most natural one is 𝑞′ ∈ 𝑉𝑆𝑞 , that means full replenishment of spent resources). So it would be necessary to develop such techniques of representation of logic of internal order ′ 𝑞 construction and completion which, from one side, would provide it’s compliance with an order 𝑞, and, from another one, would provide replenishment of namely power storages, applied for an order 𝑞 completion, by consumption of any other resources having place in an EI resource base. Such an approach fully fits the reality, where PSs are recharged on a regular basis by consumption of other resources entering an EI RB. Thus, firstly, a presumption that an EI is a closed system would be satisfied, and, secondly, all power storages, applied whilst order 𝑞 completion, would be recharged by means of only internal capabilities of an energy infrastructure. The proposed multigrammatical representation of this important feature of EIs and their fragments is as follows. Let 𝑣 ∈ 𝑉𝑆𝑞 is a collection of resources consumed for an order 𝑞 completion, so 𝑣 ⊆ 𝑣𝐸 . We shall define a submultiset 𝑣′ of a multiset 𝑣 in such a way that multiobjects entering 𝑣′ represent amounts of power delivered from PSs during an order 𝑞 completion (all such multiobjects, obviously, have a form 𝑛 ∙ (𝒌𝑾𝒉: 𝑥) ), so 𝑣′ = { 𝑛 ∙ (𝒌𝑾𝒉: 𝑥) | 𝑛 ∙ (𝒌𝑾𝒉: 𝑥) ∈ 𝑣} (100) Namely these amounts of electric power would be replenished in power storages before the next order would income to an EI, so this multiset is nothing but a needed internal order, i.e., for the first glance, 𝑣′ = 𝑞 ′. (101) however, the substantial difficulty, breaking (101), is that to be an order, completed by some chain of energy transfers and transmissions, an MS 𝑣′ in a general case would contain non-terminal objects, i.e. objects, being headers of unitary rules, representing an EI. At the same time an MS 𝑣′, being a submultiset of an MS 𝑣𝐸 , contains only terminal objects. To avoid this deadlock, we propose the following solution. To define logic of PSs replenishment (recharge) a set 𝑅𝐸 would contain URs of a form (∗ 𝒌𝑾𝒉: 𝑥) → 𝑋, (102) where bold symbol " ∗ " means that to replenish one kilowatt∙hour at PS, located at a place 𝑥, it is sufficient to complete an order being a set, containing all multiobjects, entering a body 𝑋. So all URs like (102), in fact, define logic of PSs replenishment. If so, then, evidently, 𝑞′ = { 𝑛 ∙ (∗ 𝒌𝑾𝒉: 𝑥) | 𝑛 ∙ (𝒌𝑾𝒉: 𝑥) ∈ 𝑣′}, (103) so after an initial order 𝑞 completion, which results in delivery of determined by 𝑞 amounts of electric power, an internal order 𝑞′ is completed, resulting in replenishment of power storages, applied during 𝑞 completion. As may be seen, power storages, applied during an internal order 𝑞′ completion, are not replenished; otherwise a process of replenishment may become recursive and too complicated in an implementation. From the practical point of view this is quite natural. Let us note, as a conclusion of this Section, that not all power storages entering a considered energy infrastructure are rechargeable (i.e. in a multigrammatical representation of an EI not all URs with headers (𝒌𝑾𝒉: 𝑥) are supplemented by URs (102) with headers (∗ 𝒌𝑾𝒉: 𝑥)); all the rest PSs are presumed of a single use, so they may be replaced after consumption of all initially accumulated power. 9. Conclusion As it was mentioned above, a criterial base, introduced in the Section 6, provides an assessment of vulnerability of energy infrastructures to destructive impacts: if an EI and an impact satisfy formulated conditions, then an EI is vulnerable; but if the aforementioned conditions are not satisfied, then it does not mean that an affected EI is substantially resilient to an impact. Let us underline, that we do not confirm, that in the case 𝑉𝑆𝑞 ≠ {∅} an EI 𝐸 is capable to complete an order 𝑞, because in a general case there would be also assessed time delays associated with production and delivery of materiel resources (though regarding ElcI and electric power, circulating via it’s networks and grids, in a general case such delays may be ignored). So, if an order includes a deadline for delivery of all necessary resources to external consumers being a source of this order, then, despite amounts of resources available for an order completion may be sufficient for this objective, even an optimal schedule of an order completion may not provide timely delivery of all necessary resources to consumers. This is an implication of non-additivity of time; time is an additive resource regarding separate device (facility), whilst regarding an EI as a whole it is non-additive because different devices may operate in parallel. So an EI, not satisfying the introduced above criteria and thus being not vulnerable in the above sense, in a general case may be not resilient to an impact. By this reason, if in a general case an order includes restrictions on duration of it’s completion, then to assess EI resilience it would be necessary to apply more general mathematical tools than unitary multiset grammars. Such tools named temporal multiset grammars were for the first time announced in [3], and their application to the assessment of EIs resilience to destructive impacts will be considered in the future publications. There is also an inverse task to be solved – namely, given a remained part of a considered EI segment and resources, to assess, what maximal subset of a full set of consumers may be provided by power and fuel in accordance with their demand. Another variation of this task is to assess whether some predefined part of consumers may be provided by power and fuel according to their demand while all the rest consumers may be provided by some part of their demand not less than some threshold values. As well, in the future publications the aforementioned in the Section 7 task of optimal recovery of an affected EI will be considered, and also a task of an assessment of “the best” part of an order which may be completed given remained after an impact part of an EI (both with and without reserve). The next extremely important task to be considered in future is a priori design of EIs being maximally resilient to the most expected sets (sequences) of impacts and consuming for a recovery minimally possible amounts of resources. Let us note, that just the same techniques as described above in this paper regarding energy infrastructures may be applied also to heating systems, heating and cooling systems, combined heat and power systems [35, 36] and water supply systems [37-39]. An application of the MGF to these systems as well as to a sewer-mining [40] is described in short in [4]. All such partial applications would be joined in the near future to the integrated application of the multigrammatical framework to the area of resilience and recovery of critical infrastructures. Acknowledgements Author is grateful to Acad. Igor Bychkov, Acad. Yuriy Shokin, and Prof. Fred Roberts for useful discussions and support. References [1] Sheremet I A 2010 Recursive Multisets and their Applications (Moscow, Nauka) p 292 (In Russian) [2] Sheremet I A 2011 Recursive Multisets and their Applications (Berlin, NG Verlag) p 249 [3] Sheremet I A 2018 Multiset Analysis of Consequences of Natural Disasters Impacts on Large-Scale Industrial Systems Data Science Journal 17(4) 1–17 [4] Sheremet I 2019 Multiset-based assessment of resilience of sociotechnological systems to natural hazards Natural Hazards Risks, Exposure, Response, and Resilience (London: IntechOpen). Available at: https://www.intechopen.com/books/natural-hazards-risk-exposure-response-and-resilience/multiset-based- assessment-of-resilience-of-sociotechnological-systems-to-natural-hazards (accessed: 10.06.2021) [5] Sheremet I 2019 Multiset-Based Knowledge Representation for the Assessment and Optimization of Large- Scale Sociotechnical Systems Enhanced Expert Systems (London: Intech Open) Available at: https://www.intechopen.com/books/enhanced-expert-systems/multiset-based-knowledge-representation-for- the-assessment-and-optimization-of-large-scale-sociotech (accessed: 10.06.2021) [6] Sheremet I 2019 Unitary Multiset Grammars and Metagrammars Algorithmics and Applications Enhanced Expert Systems (London: Intech Open). Available at: https://www.intechopen.com/books/enhanced-expert- systems/unitary-multiset-grammars-an-metagrammars-algorithmics-and-application [7] Sheremet I A 2021 Application of the Multigrammatical Framework to the Assessment of Resilience and Recoverability of Large-Scale Industrial Systems Resilience in the Digital Age. Lecture Notes in Computer Science (Springer) 12660 pp 16–34 [8] Wallace M 2002 Constraint logic programming Computational Logic: Logic Programming and Beyond. Lecture Notes in Computer Science (New York: Springer) 2407 pp 512–556 [9] Apt K 2003 Principles of Constraint Programming (Cambridge: Cambridge University Press) p 420 [10] Frunkwirth T, Abdennadher S 2003 Essentials of Constraint Programming (Berlin: Springer Verlag) p 398 [11] Kendal S L and Green M 2007 An Introduction to Knowledge Engineering (London: Springer) p 300 [12] Akerkar R and Sajja P 2010 Knowledge-Based Systems (Sudbury: Jones and Bartlett Publishers) p 350 [13] Bratko I 2012 Prolog Programming for Artificial Intelligence (New York: Addison Wesley) p 696 [14] Karlot J K 2005 Integer Programming: Theory and Practice (CRC Press) p 336 [15] Roberts F S 2009 Applied Combinatorics (London: Chapman & Hall / CRC) p 860 [16] Lasdon S L 2013 Optimization Theory for Large Systems (New York: Dover Publications) p 560 [17] Taha H A 2016 Operations Research: An Introduction (London: Pearson) p 838 [18] Sheremet I 2020 Resource-based games Proceeding of the 2nd International Workshop on Information, Computation, and Control Systems for Distributed Environments (Irkutsk, Russia) 2638 pp 234–251 [19] Sheremet I 2020 Multigrammatical Approach to the Assessment of Sustainability of Intelligent Transport Systems 2020 Proceeding of the International Conference on Information Technology and Nanotechnology (Samara, Russia) pp 1–6 [20] Rinaldi S, Peerenboom J P and Kelly T K 2001 Identifying, understanding, and analyzing critical infrastructure interdependencies IEEE Control Systems Magazine 21 pp 11–25 [21] Katay M E 2010 Electric power industry as critical infrastructure Network World. Available at: https://www.networkworld.com/article/2217677/datacenter/electric-power-industry-as-critical- infrastructure.html [22] Nepal R and Jamasb T 2013 Security of European electricity systems: conceptualizing the assessment criteria and core indicators International Journal of Critical Infrastructure Protection 6(3-4) pp 182–196 [23] Larsen sen E R, Osorio S and van Ackere A 2017 A framework to evaluate security of supply in the electricity sector Renewable and Sustainable Energy Reviews 79 646–655 [24] He X and Cha E J 2018 Modeling the damage and recovery of interdependent critical infrastructure systems from natural hazards Reliability Engineering and System Safety 177 pp 162–175 [25] Liu K, Wang M, Zhu W, Wu J and Yan X 2018 Vulnerability analysis of an urban gas pipeline network considering pipeline road dependency International Journal of Critical Infrastructure Protection 22 pp 125– 138 [26] Rechak D, Senovsky P, Hromada M and Lovecek T 2019 Complex approach to assessing resilience of critical infrastructure elements International Journal of Critical Infrastructure Protection 25 pp 125–138 [27] Stergiopoulos G, Kotzanikolaou P, Theocharidou M, Lykou G and Gritzalis D 2016 Time-based critical infrastructure dependency analysis for large-scale and cross-sectoral failures International Journal of Critical Infrastructure Protection 12 pp 46–60 [28] Chomsky N 2005 Syntactic Structures (The Hague: Mouton de Gruyter) p 159 [29] Hopcroft J E, Motwani R and Ullman J 2001 Introduction to Automata Theory, Languages, and Computation (Reading, Mass.: Addison–Wesley) p 528 [30] Jesse B-J, Heinrichs H U and Kuckshinrichs 2019 Adapting the theory of resilience to energy systems: a review and outlook Energy, Sustainability and Society 9(27) pp 1–19 [31] Wang J, Zuo W, Rhode-Barbagidos L, Lu X, Wang J and Lin Y 2019 Literature Review on Modeling and Simulation of Energy Infrastructures from a Resilience Perspective Reliability Engineering and System Safety 183 pp 360–373 [32] Ji C, Wei Y and Poor H V 2017 Resilience of Energy Infrastructure and Services: Modeling, Data Analytics, and Metrics Proceedings of the IEEE 105(7) pp 1354–1366 [33] Brown R E 2008 Electric Power Distribution Reliability (Boca Raton, FL: CRC Press) p 110 [34] Lavaei J, Tse D and Zhang B 2014 Geometry of power flows and optimization in distributed networks IEEE Transactions on Power Systems 29(2) pp 572–583 [35] Werner S 2017 International review of district heating and cooling Energy 137 pp 617–631 [36] Mazher A R, Liu S and Shukla A 2018 A state of art review on the district heating systems Renewable and Sustainable Energy Reviews 96 pp 420–439 [37] Kurian M and McCarney P 2010 Peri-urban Water and Sanitation Services. Policy, Planning and Method (New York: Springer) p 300 [38] Water supply. Available at: www.who.int.read/em2002chap7 (accessed: 10.06.2021) [39] Water supply system. Encyclopedia Britannica. Available at: https://www.britannica.com/technology/water- supply-system (accessed: 10.06.2021) [40] Makropoulos C, Rozos E, Tsoukalas I, Plevri A, Karakatsanis L and Karagiannidis L 2018 Sewer-mining: A water reuse option supporting circular economy, public service provision and entrepreneurship Journal of Environmental Management 216 pp 285–298