Electrical Power and Energy Systems 61 (2014) 192–196

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

Modified fuzzy algorithm based safety analysis of nuclear energy for

sustainable hydrogen production in climate change prevention
Tae Ho Woo ⇑
Systemix Global Co. Ltd., 3 F, 494-48, Yonggang-dong, Mapo-Gu, Seoul 121-876, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: The high temperature coolant of nuclear power plants (NPPs) has been investigated for the hydrogen
Received 20 November 2012 production, which could be a major role of a green energy promotion. An accident of the high temperature
Received in revised form 29 October 2013 gas cooled reactor (HTGR) is modeled for the stabilized hydrogen production using nuclear energy. For the
Accepted 20 March 2014
clean energy resource pursuit in preventing the climate change, the hydrogen is one of very attractive
energy sources. The non-operational data could be produced by the fuzzy set theory which is one of
non-linear complex algorithms. So, the result can show the possibility of the event happening instead
Keywords:
of the exact solutions. The random numbers are generated for membership numbers of the fuzzy function.
Hydrogen
Nuclear power plants
The event manipulation is done by new membership numbers for the propagations. The final result is 1.0
Gas in 8 times during 100 months. So, the frequency is 0.08, or 8% of successful long-term cooling by
Random number conduction.
Fuzzy set theory Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction step by step [4]. The reactions are decomposing water thermally
into hydrogen and oxygen using heat at temperatures lower than
The high temperature gas cooled reactor (HTGR) has been 900 °C, which can be made by HTGR. According to the gas phase
suggested for the hydrogen production as well as the electricity decomposition theory, the operational pressure could be
generation in order to cope with the carbon dioxide matter in maintained as lower value, although the capability of heat exchan-
the aspect of global warming as climate change [1–3]. Until now, ger increases by the pressure [5].
hydrogen is being used as raw materials of chemical products such In the historical review of the HTGR, the design was first pro-
as ammonia, but hardly used as an energy carrier like electricity posed by the staff of the Power Pile Division of the Clinton Labora-
[4]. So, the high temperature of the plant coolant can produce tories (known now as Oak Ridge National Laboratory) in 1947 [6].
the hydrogen with the electricity generation. There are technical The Peach Bottom reactor was the first HTGR to produce electricity
problems to be commercialized. In the case of Fort St. Vrain with operation from 1966 through 1974 as a technology demon-
Generating Station, it was stopped by some problems of economic strator. In addition, Fort St. Vrain Generating Station was one
factors, which was operated from 1979 to 1989 [3]. After this time, example of this design that operated as an HTGR from 1979 to
there are no commercial HTGR and some HTGRs exist as Dragon 1989. However, the reactor was remodeled by some problems to
reactor (United Kingdom), AVR reactor and THTR-300 (Germany), its decommissioning due to economic factors. This was proved as
HTTR (Japan), and HTR-10 (China) [3]. It is expected that the the HTGR concept in the United States. In other countries, there
hydrogen will be widely used as clean energy carrier for fuel cells was the Dragon reactor in the United Kingdom, AVR (Arbeitsgeme-
to generate electricity, because hydrogen can make a major role to inschaft Versuchsreaktor in German) and Thorium high tempera-
reduce carbon dioxide which can produce the green house effect. ture reactor-300 (THTR-300) in Germany, the High temperature
Therefore, the global R&D has focused on the fuel cell vehicles test reactor (HTTR) in Japan, and the High temperature reactor-
(FCV) and stationary power generators. There are equations for 10 (HTR-10) in China. China has constructed the 2 pebble-bed
hydrogen production by chemical reactions. As for the hydrogen HTGRs. There are several descriptions of the HTGR. The specifica-
production technology, a thermo-chemical hydrogen production tions of the HTGR are in Table 1, which is a gas turbine modular
cycle so-called IS (iodine–sulphur) process has been developed helium reactor (GT-MHR) [7].
For the stable sustainable hydrogen productions, the accident
⇑ Tel.: +82 2 831 8394; fax: +82 2 831 8422. assessment is investigated in the very high temperature reactor
E-mail address: [email protected] (VHTR). The fuzzy set theory is applied to the safety assessment

http://dx.doi.org/10.1016/j.ijepes.2014.03.035
0142-0615/Ó 2014 Elsevier Ltd. All rights reserved.
 T.H. Woo / Electrical Power and Energy Systems 61 (2014) 192–196 193

Nomenclature

ATWS anticipated transient without scram HTTR high temperature test reactor
AVR Arbeitsgemeinschaft Versuchsreaktor ICBM Intercontinental Ballistic Missile
CFR Code of Federal Regulation NPPs nuclear power plants
FAA Federal Aviation Administration NRC Nuclear Regulatory Commission
FCV fuel cell vehicles THTR-300 Thorium high temperature reactor-300
GT-MHR gas turbine modular helium reactor VHTR very high temperature reactor
HTGR high temperature gas cooled reactor

in this advanced power plant of the nuclear power plants (NPPs). using the nuclear fuel. That is to say, although the 1600 °C is con-
The conventional method is modified by the fuzzy set logic. The sidered as the maximum temperature in the nuclear fuel, there is
safety assessment has the event/fault tree concept for the analysis the maximum event failure frequency at TFuel = 1250 °C.
which has been used for the license of the plant constructions and
operations. Since there are no operational data in the HTGR, 2.2. Event/fault tree analysis
non-linear algorithm like the fuzzy set theory of the artificial
intelligence methods could make the quantifications. Language Fault Tree Analysis (FTA) was created in 1962 at Bell Laborato-
verification is used in the fuzzy algorithm, where the new kind of ries by H.A. Watson, which was done by a U.S. Air Force Ballistics
the function as the membership function incorporated with the Systems Division contract to evaluate the Minuteman I Interconti-
membership number in the theory is utilized. nental Ballistic Missile (ICBM) Launch Control System [12–14].
In the literature review, Ball studied the results of various Boeing Company and Avco Corporation expanded usage of FTA to
accident scenario simulations for the two major modular HTGR the entire Minuteman II system in 1963–1964. In 1970, the U.S.
variants (prismatic and pebble bed cores) are presented [8]. Federal Aviation Administration (FAA) published a change to 14
Sensitivity studies can help to quantify the uncertainty ranges of CFR (Code of Federal Regulation) 25.1309 airworthiness regula-
the predicted outcomes for variations in some of the more crucial tions for transport category aircraft in the Federal Register at 35
system parameters, as well as for occurrences of equipment and/or FR 5665.
operator failures or errors. In addition, safety demonstration tests The U.S. Nuclear Regulatory Commission (NRC) began using
using the HTTR will be conducted for the purpose of demonstrating probabilistic risk assessment (PRA) methods including FTA in
inherent safety features of HTGRs as well as providing the core and 1975 of the nuclear power industry, and significantly expanded
plant transient data for validation of HTGR safety analysis codes PRA research following the 1979 incident at Three Mile Island
[9]. Also, safety demonstration tests were conducted on the [15]. This led to the 1981 publication of the NRC Fault Tree Hand-
10 MW HTR-10 to verify the inherent safety characteristics of book NUREG-0492 [16], and mandatory use of PRA under the NRC’s
modular HTGRs as well as to obtain the transient data of reactor regulatory authority.
core and primary cooling system for validation of HTGR safety There are successful works for nuclear power age as power
analysis models and codes [10]. reactor of the enhanced safety in which the advanced safety crite-
ria with conservativeness as new kinds of safety measures and the
consistent analysis in the fields of general nuclear energy and
2. Method
human health matters had been done in late 1960s [17]. In the
early 1970s, the US Atomic Energy Commission started the safety
2.1. Modeling
of NPPs by the Reactor Safety Study ‘WASH-1400’ which ended
in October 1975 [18].
For the stabilized hydrogen production, the accident is modeled
Fig. 1 shows the conventional event/fault tree method where
for the assessment, which can break the hydrogen generations. The
the event flows from left side to right side in event tree (a). The
anticipated transient without scram (ATWS) is one of important
fault tree is shown in (b). The models sequences of events that
accidents in the HTGR. The safety criteria for ATWS and its
need to occur in order for undesired end states to occur. A
parameter are shown in the ‘Screening of Gas-Cooled Reactor
sequence of events means an accident sequence. The example of
Thermal–Hydraulic and Safety Analysis Tools and Experiment
an accident sequence is a fire that leads to catastrophic
Database’, I-NERI Final Project Technical Report, Korea Atomic
consequences because mitigation systems fail to operate. A model
Energy Research Institute, Korea [11]. Table 2 is the key events of
ATWS. These values are used to find out the event failure frequency
Table 2
Scenario of ATWS.
Table 1
Specifications of the GT-MHR. Step Event scenario
1 Rod withdrawal and scram failure
Content Value
2 Shutdown cooling system fails to start
Reactor power (MWt) 600 3 Core power increases by reactivity insertion
Tin/Tout (°C) 491/850 4 Core temperature and system pressure increase
Number of fuel blocks/pebbles 1020 5 Power control by runback
Bypass flow fraction (%) 10–15 6 Reactor trip signal and turbine trip, but, no reactor scram
Reactor pressure (bars) 70 7 Coastdown of primary flow
Power density (W/cc) 5 8 PCM to rapidly heatup core
Reactor mass flow rate (kg/s) 320 9 Core power decreases by Doppler feedback, Xenon inventory increases
Effective core height (m) 7.93 10 Equilibrium power level is to be decay heat
Core diameter (m) 2.63 ID/4.83 OD 11 Long-term conduction and radiation cooling
Reactor pressure (bars) 70 12 Re-criticality and power oscillation
194 T.H. Woo / Electrical Power and Energy Systems 61 (2014) 192–196

(a) A 1.0

Membership number
B

C Interested interval of
membership number
(b)
A

Fig. 2. New assessment by fuzzy set theory.

The necessity of the modified theory is due to the non-experi-

ence or insufficient data of the analysis. In HTGR, there have been
B C
nearly no commercial operations. The fuzzy set could be useful to
construct the assessment tool by its non-linearity. In the fuzzy set
Fig. 1. Conventional event/fault tree (a) event tree, (b) fault tree. theory, the Fig. 2 shows the membership function has the inter-
ested interval of the membership number. In the interval values,
of a simple event sequence in a PRA is shown in Fig. 1(a). The
the random numbers are obtained. Then, the propagations are cal-
intrinsic to a fault tree is the concept that an outcome is a binary
culated by plus or multiplications. Therefore, the modified logic
event i.e., to either success or failure [19]. The fault tree is
shows the algorithm as follows,
composed of a complex of entities known as gates that serve to
permit or inhibit the passage of fault logic up the tree. Gates mean
If a < random number R1 < b, then R1 selected. Otherwise, next
the relationships of events needed for the occurrence of an upper
random number is selected,
event. The upper event is the output of the gate. Hence, the lower
If a < random number R2 < b, then R2 selected. Otherwise, next
events are the input data to the gate. The gate symbol denotes the
random number is selected,
type of relationship of the input events required for the output
If a < random number R3 < b, then R3 selected. Otherwise, next
event, which is seen in Fig. 1(b).
random number is selected.

2.3. Modified fuzzy analysis

For propagation, R1R2R3 is calculated for event flow.
For fault tree, there are two cases as follows;
Fuzzy logic was introduced in 1965 by Dr. Lofti A Zadeh in his
paper of fuzzy sets [20]. Dr. Zadeh and others continued to develop
AND gate case: R1R2R3
fuzzy logic at that time. Since some of the underlying mathematics
OR gate case: R1 + R2 + R3
had not yet been explored, the idea of fuzzy sets and fuzzy logic
were not accepted well within academic societies. The application
The fuzzy set function is constructed by the expert’s judgment,
of fuzzy set theory had been developed in 1960s and 1970s when
which is a definition of the fuzziness in the theory. So, the interval
the 1st successful practical product is control related stuffs like the
of the membership number is decided by the expert’s judgment.
heating system and cement factory system [21]. Then, the Japanese
The fuzziness means that something to be difficult to perceive
applications had manufactured in 1980s washing machines, video
[25] in which the vague understanding of event is implied. So, the
cameras, cranes, subway trains, and so on and, eventually, the
probability could be interpreted by the fuzziness, because there is
applications returned to the United States in the late 1980s [21].
a common property of non-exact-expectation in future event.
Hitachi used a fuzzy controller for the Sendai train control,
Hence, the PRA could be substituted by fuzzy logic. When the appli-
which used an innovative system created by man in 1987 and
cation of fuzziness in risk analysis is performed, the proposed condi-
the controller has been doing his job very efficiently [22,23]. In
tions are used in this paper where the random numbers with
1993, Fuji Fuzzy Logic applied to control chemical injection water
specified intervals are used. That is, the constrained condition can
treatment plants for the first time and then there were some more
select the optimized value where a random number among R1, R2,
height Fuzzy Logic has been used in government, universities and
and R3 is selected with if-then condition. Then, the AND gate and
industries [22,23].
OR gate are calculated by the same manner to the conventional tree
The main meaning to fuzzy systems is that truth values (in fuzzy
method. The expert means the judgment of the operator or its equiv-
logic) or membership values (in fuzzy sets) are described with the
alent person who can make a decision of event value which has limit
value of range [0.0, 1.0], with 0.0 showing the Falseness and 1.0
of 1 or 0 where the 1 is success and 0 is failure of the designed event.
showing the Truth [24]. For example, using the references [24], if
Defuzzification is not used, because the final values are compared
one takes the statement, ‘Tom is old’. If Tom is 77 years old, one
each other as the comparative value without dimension.
could mention a statement of the truth value as 0.80. The sentence
is regarded as ‘Tom is a member of the set of old people’ [24]. So,
this statement would be expressed by fuzzy sets as follows, 3. Calculation and result
mAGEðTomÞ ¼ 0:80
The scenario is described as following 5 steps, which is a
where m is the membership function for the fuzzy set of old people contraction of the Table 2. That is, this means the event flows by
with a value between 0.0 and 1.0 [24]. sequences of the scenario as follows,

Flow Coastdown Long-Term

SCS Fail to Start Pre-Turbine Trip Recriticality Cooling by
& Power Eq.
Conduction

Fig. 3. Modified event tree for long-term cooling by conduction.

 T.H. Woo / Electrical Power and Energy Systems 61 (2014) 192–196 195

Table 3 production aspect. The nuclear energy could be dangerous, if the

Values for membership number interval. severe accident happened. However, the perfect control of the al-
Event Value most all NPPs operating countries can produce the green and clean
Shutdown cooling system start fails 0.3 < random number < 0.7 hydrogen energy with mass production level. For the clean energy
Trip of pre-turbine 0.5 < random number < 0.9 pursuit, the hydrogen is one of very attractive energy sources.
Flow coast down and power equilibrium 0.5 < random number < 0.9 Hence, the nuclear energy based hydrogen production is very effec-
Re-criticality achieved 0.5 < random number < 0.9 tive method, because it is easy to make the high temperature to ex-
Long-term cooling by conduction 0.3 < random number < 0.6
tract the hydrogen from the water.
The aleatory of fuzziness is explained that since the temporal
indifference is needed in the aleatory interpretation of probability
Long-Term Cooling by Conduction from the randomness in empirical phenomena, the non-empirical
1 event could use the concept of fuzziness in the uncertainty like the
nuclear accident [26]. There are some methods used in propagating
epistemic uncertainties as interval analysis, Dempster–Shafer
0.75 evidence theory, and second-order probability where the examples
are regarding structural dynamics [27]. The non-operated HTGR
0.5
needs this fuzzy algorithm due to the uncertainty expression as
the realizations. The quantification of the uncertainty could be effec-
tively shown in the fuzzy set theory comparing to the conventional
0.25 probability where the restriction of the height of the probability
distribution of 1.0 can be changed by the operator’s subjective logic
as the membership number. In addition, the distribution of graph
0 can be changed as the several kinds of geometries as triangular,
0 10 20 30 40 50 60 70 80 90 100 rectangular, or circular. Therefore, the fuzzy set logic can be used
Time (Month) successfully in the non-experienced data like the case of HTGR.
"Long-Term Cooling by Conduction" : Current In this paper, one accident of the VHTR is investigated for the
Fig. 4. Result of assessment for long-term cooling by conduction.
stabilized hydrogen production using nuclear energy. The
modeling of the accident assessment of the advanced NPPs has
been performed. The fuzzy set logic is a major algorithm which
(1) Shutdown cooling system start fails. is substituted with the conventional event/fault tree method. There
(2) Trip of pre-turbine. are several significant important key matters in this study.
(3) Flow coast down and power equilibrium.
(4) Re-criticality achieved.  Hydrogen is produced by the nuclear energy.
(5) Long-term cooling by conduction.  The non-linear algorithm is used for the accident
assessment.
The Fig. 3 shows a modified event tree for long-term cooling by  The energy production is analyzed by the data
conduction where the event scenario is described. The values are quantifications.
obtained by the random sampling which is shown in Table 3. The  Safety in NPPs could be enhanced by the artificial intelli-
scenario is simplified from the Table 2 where the full scenario of gence analysis.
ATWS is explained. Hence, this simplified scenario can show the  The quantification of the propagation is performed by the
calculation sequence of the newly introduced modified fuzzy random number with membership numbers.
algorithm. If the values are in the interval, the representative value
is 1.0. Otherwise, it is 0.0. Then, all representative values are There are many kinds of the applications for the nuclear energy
multiplied. The final result is 1.0 in 8 times during 100 months, based hydrogen productions. The hydrogen car can make the air
which is in Fig. 4, where the axis is dimensionless because the quality of the city be cleaner where the air pollution is one of se-
success is shown as 1.0 and the failure is as 0.0. So, the frequency vere urban matters. The remaining matter of the car is just water
is 0.08, or 8% of successful long-term cooling by conduction. after combustion of the hydrogen in the car engine. In addition,
The calculation is used for the modified scenario where the the mass energy production is possible using VHTR comparing to
random number based risk analysis is quantified. In HTGR, the mer- the conventional hydrogen production methods, because the
it of safety system is that the gas coolant can cool down the heat of capacity of the NPPs is bigger where the power capacity is
nuclear fuel effectively by conduction with long-period. Therefore, 600 MWt in the GT-MHR as it is said above. In some more indus-
this scenario is used for new event tree in which the event success is tries, the other matters could be modeled by the fuzzy set theory.
explained by the long-term period cooling by the gas coolant. For example, the airplane flying could be assessed by the fuzzy set
The value of 8% of successful long-term cooling by conduction is theory for the safe operations. The airplane flying has many uncer-
quite low for the real operational system in the commercialization. tainties in the travel. So, the dangerous situations can be expressed
This result is from the membership number interval where the by the fuzzy logic with membership function.
range is so wide. In the real application, it should be optimized
by the expert’s mind where the range is narrowed. This paper
intended to show the clear application of range valued fuzziness. References
Hence, the un-real result could be possible and the conceptual
meaning is important of the applications. [1] IAEA, Current status and future development of modular high temperature gas
cooled reactor technology, 2001. http://www-pub.iaea.org/MTCD/Publications/
PDF/te_1198_prn.pdf.
4. Conclusions [2] NEA, The Very-High-Temperature Reactor (VHTR) is a graphite-moderated,
helium-cooled reactor with a thermal neutron spectrum, 2013. http://
www.gen-4.org/Technology/systems/vhtr.htm.
The new resource for making hydrogen by the high temperature [3] Wikipedia, Very high temperature reactor, 2013. http://en.wikipedia.org/wiki/
of the coolant is very important due to the carbon free energy Very_high_temperature_reactor.
196 T.H. Woo / Electrical Power and Energy Systems 61 (2014) 192–196

[4] Ogawa M, Hino R, Inagaki Y, Kunitomi K, Onuki K, Takegami H. Present status [17] Keller W, Modarres M. A historical overview of probabilistic risk assessment
of HTGR and hydrogen production development in JAEA. OECD/NEA, 2010. development and its use in the nuclear power industry: a tribute to the late
[5] Rodgers MC. Summary report on design and development of high temperature Professor Norman Carl Rasmussen. Reliab Eng Syst Saf 2005;89(3):271–85.
gas-cooled power pile. TN, USA: Oak Ridge; 2009. [18] Haynes M. The evolution of probabilistic risk assessment in the nuclear
[6] General Atomics, 2012. http://gt-mhr.ga.com/. industry. Process Saf Environ Prot 1999;77(3):117–42.
[7] Ball S. Sensitivity studies of modular high-temperature gas-cooled reactor [19] Bryan K. http://www.calvin.edu/~pribeiro/othrlnks/Fuzzy/home.htm. Calvin
postulated accidents. Nucl Eng Des 2006;236:454–62. College Engineering Department. Calvin College CIT, 2012.
[8] Tachibana Y, Nakagawa S, Takeda T, Saikusa A, Furusawa T, Takamatsu K, et al. [20] Bai Y, Zhunang H, Wang D. Advanced fuzzy logic technologies in industrial
Plan for first phase of safety demonstration tests of the High Temperature applications. Springer; 2006.
Engineering Test Reactor (HTTR). Nucl Eng Des 2003;224:179–97. [21] Zimmermann H. Fuzzy set theory. WIREs Comput Statist 2010;2:317–32.
[9] Chen F, Dong Y, Zhang Z, Zheng Y, Shi L, Hu S. Post-test analysis of helium [22] Sugeno M. Fuzzy measures and fuzzy integrals: a survey (Gupta MM, Saridis
circulator trip without scram at 3 MW power level on the HTR-10. Nucl Eng GN, Gaines BR, editors), Fuzzy Automata and Decision Processes, North-
Des 2009;239:1010–8. Holland, NY, 1977. p. 89–102.; Re-cited Garrido A, A Brief History of Fuzzy
[10] KAERI. Experiment Database: I-NERI Final Project Technical Report. Korea Logic, BRAIN (Broad Research in Artificial Intelligence and Neuroscience), vol.
Atomic Energy Research Institute (KAERI), Republic of Korea, 2007. 3(1), 2012.
[11] Ericson C. Fault tree analysis – a history. In: Proc of the 17th International [23] Sugeno M. Industrial applications of fuzzy control, Elsevier Science Pub. Co.;
Systems Safety Conference, 1999. Re-cited Garrido A, A Brief History of Fuzzy Logic, BRAIN (Broad Research in
[12] Rechard R. Historical relationship between performance assessment for Artificial Intelligence and Neuroscience), vol. 3(1), 2012.
radioactive waste disposal and other types of risk assessment in the United [24] Brule JF. Fuzzy systems – a tutorial, 2005. http://www.austinlinks.com/Fuzzy/
States. Risk Anal 1999;19:763–807. tutorial. Html.
[13] Benner L. Accident theory and accident investigation. In: Proc 1975 the Society [25] Oxford dictionaries, Oxford University, http://www.oxforddictionaries.com/
of Air Safety Investigators Annual, Seminar, 1975. definition/english/fuzzy, Oxford dictionaries, 2013.
[14] US NRC. Severe Accident Risks, An Assessment for Five U.S. Nuclear Power [26] Bernardini A, Tonon F. Aggregation of evidence from random and fuzzy sets.
Plants. Washington, DC: NUREG-1150, U.S. Nuclear Regulatory Commission; Zeitschrift für Angewandte Mathematik und Mechanik J Appl Math Mech
1990. 2004;84(10–11):700–9.
[15] US NRC. Fault Tree Handbook. Nuclear Regulatory Commission. NUREG-0492, [27] Swiler P, Paez L, Mayes L. Epistemic uncertainty quantification tutorial. Orlando,
1981. Florida USA: Proceedings of the IMAC-XXVII; 2009. February 9–12.
[16] NASA. Fault tree handbook with aerospace applications. NASA Office of Safety
and Mission Assurance NASA Headquarters Washington, DC, 2002.