TEPSYS developed a BWR loading pattern(LP) optimization system FINELOAD-3 in 1990 as a prototype. In TEPSYS, FINELOAD-3 has been applied for actual BWR reload core design works since 1998 for TEPCO plants because of dramatic evolution of computer speed. FINELOAD-3 has been confirmed its high performance for reload core design works by other utilities and has been installed ten utilities and fuel vendors in Japan, Europe and US. In order to reduce refueling shuffling work in the reactor outage, TEPSYS and some utilities have applied FINELOAD-3 minimum shuffling option in the core design work.

FINELOAD-3 works in combination with an advanced nodal code SIMULATE-3, which is developed by Studsvik Scandpower Inc. described in reference [8], as core simulator or with other core simulators. FINELOAD-3 employs a simple linear perturbation (SLP) method and a modified Tabu search method to develop an optimal LP, which the initial LP and a control rod pattern (CRP) sequence are given. The objective function is to maximize the end of cycle (EOC) core reactivity while satisfying all specified constraints (i.e., Minimum critical power ratio (MCPR) and maximum linear heat generation rate (MLHGR) as well as Shutdown Margin (SDM) limit). FINELOAD-3 has a capability of adjusting the CRP at each depletion point applicable for many CRP strategies.

In this paper, the definition of “optimized” is that core characteristics of LP is equivalent to or better than that of LP designed by skilled engineers manually.

(1)System Outline
The flowchart of the FINELOAD-3 system outline is shown in Fig.1. Four steps are involved in this optimization system.

Step-1
The first step is the selection of the initial guess LP and CRP sequence. The fresh assembly locations and CRP sequence should be determined empirically using reasonable judgment by user. Because FINELOAD-3 does not change the fresh assembly locations and CRP sequence. The initial location of depleted fuel assemblies is not critical. However reasonable selection of peripheral fuel assemblies will reduce the computer time during the LP optimization process.

Step-2
The second step is the generation of a reference solution (i.e., cycle depletion and thermal limits calculation at each depletion step and SDM evaluation at several depletion steps) using core simulator (i.e., SIMULATE-3) with fine mesh (i.e., twenty-four axial nodes).

Step-3
The third step is the search for an optimized LP using FINELOAD-3. Fuel assembly binary swaps and core characteristics estimations using a SLP method are repeated until a LP which satisfies all the constraints is obtained or the iteration number reaches the specified number (e.g., about fifty to sixty).

Step-4
The forth step is the CRP adjustments for the best LP (not final LP) obtained in FINELOAD-3. The adjusted CRP sequence and the best LP are passed to core simulator for direct verification and the reference solution is updated.

The optimization process repeats between Step-2 and Step-4 until the iteration number (NS) reaches the specified limit number, NSmax.

Fig. 1The flowchart of FINELOAD-3 system
(The detail methodology of FINELOAD-3 is described in Reference [1].)

1. (1) Search Space :
   - Optimization : Depleted assembly locations (Fresh assembly locations are fixed by user.)
   - Core/CR Symmetry : Half, Quarter(Mirror or Rotation), Octant
   - Search Mode : Simultaneous LP and CRP search, LP search with fixed CRP, CRP search with fixed LP
   - Flexibility Search : Fixing some depleted assembly locations by user input. Fixing some positions of CRP at any depletion steps by user input
   - CRP Strategy : Control Cell Core (CCC), Conventional

1. (2) Constraints
   - Thermal Margin Limits : MCPR, MLHGR, maximum average planner linear heat generation rate (MAPLHGR)
   - Assembly average relative power fraction
   - SDM
   - Loading Constraints for each fuel location
   - Maximum burn-up limits : Assembly average, Rod average, Pellet
   

1. (1)Decrease fuel cycle cost by reducing fresh assembly batch size due to high optimized LP level The optimized LP level obtained by FINELOAD-3 is equivalent to or better than LPs optimized by skilled engineer manually. FINELOAD-3 can reduce the fresh assembly batch size about 0 – 8 assemblies compared with manually optimized LP in a single cycle reload core design whose the total number of fuel assemblies is 764 assemblies for 14 month operation.

1. (2) Man power saving and wider search space User can search multiple strategies such as several fresh assembly locations, several fresh assembly batch sizes, several CRP strategies etc., in parallel using several CPUs. Therefore, FINELOAD-3 user can search wider space and find higher optimized LPs compared with manually optimized LP in the same working days.

1. (3) Man power saving and wider search space User can search multiple strategies such as several fresh assembly locations, several fresh assembly batch sizes, several CRP strategies etc., in parallel using several CPUs. Therefore, FINELOAD-3 user can search wider space and find higher optimized LPs compared with manually optimized LP in the same working days.

Table. 1 An example of computational time for LP optimization

^*1 S3: SIMULATE-3 *2 FL3:FINELOAD-3

Manual LP optimization by skilled engineers usually takes about 1 or 2 weeks.

1. (4) Options of FINELOAD-3
   a) Equilibrium core development option
   - Develop an equilibrium core LP or consecutive cycle core LPs using FINELOAD-3 system (Detail description is shown in the Reference [3])
   b) Minimum shuffling option
   - Search for an optimal LP while minimizing fuel relocation time with small penalty on core characteristics (Detail description is shown in the Reference [4])
   c) Partial power option
   - Search for an optimal LP which satisfies thermal limits constraints at partial power conditions as well as rated power conditions
   d) Squeeze option
   - Search for an optimal LP which yields high burnup for the discharge assemblies
   e) Pool option
   - Generate a series of initial guess LPs that cover specified range of fresh fuel type combinations

1. (1)Example of optimized LP obtained by FINELOAD-3 system
   　Reload core design condition is as follows;
   a) Total number of fuel assemblies : 548
   b) Number of fresh fuel assemblies : 128
   c) Operation length : 14 months
   d) CR strategy : Control Cell Core
   The optimized LP obtained by FINELOAD-3 system is shown in Fig.2. The values in the red box are the fraction of thermal margin against the design target in the fresh assembly locations. These values are close to 1.0. Also, the values at control rod position are SDM whose target value is more than 1.3%. Some SDM values are close to 1.3%. Therefore, the LP is well optimized by FINELOAD-3 system.

Fig.2 An example of optimized LP obtained by FINELOAD-3 system

The optimization trend for each SIMULATE-3 calculation is shown in Fig.3. The EOC core reactivity of the best LP is higher than that of the initial LP.

Fig.3 An optimization trend of FINELOAD-3 system

1. (2)Experiences
   a) Actual reload core design
   - More than 80 cycles for TEPCO BWR3,4,5 and ABWR
   - More than 10 cycles for TVO(Finland), VNF(Sweden) etc.
   b) Operation length
   - 10 – 24 Months
   c) Fuel bid evaluation
   - Comparison between GE-11 and Atrium-9 for TEPCO
   d) Fuel or Gd design study using FINELOAD-3 equilibrium core development option
   e) Minimum shuffling reload core design study using FINELOAD-3 minimum shuffling option

1. (3)User list
   Japan : TEPSYS, Tohoku Electric Power CO.(TOiNX), Japan Atomic Power CO.(GIS), Chubu Electric Power CO.(CTI), Nuclear Fuel Industries
   Finland : Teollisuuden Voima Oyj
   Sweden : Vattenfall Nuclear Fuel, OKG, Westinghouse Electric Sweden
   Spain : Iberdrola Generacion
   US/Germany : Areva
   

1. [1] S. JAGAWA, T. YOSHII and A. FUKAO, “Boiling Water Reactor Loading Pattern Optimization Using Simple Linear Perturbation and Modified Tabu Search Methods,” Nucl. Sci. Eng. 138, 67(2001)
2. [2]S. JAGAWA, “Improved Control Rod Adjustment Model In The Loading Pattern Optimization Code FINELOAD-3,” Proc. ANS Topical Meeting ANFM-III, Hilton Head Island, South Carolina (2003)
3. [3]T. YOSHII and S. JAGAWA, ”Equilibrium core loading pattern generation system using FINELOAD-3,” Trans. Am. Nucl. Soc., 90, 592(2004)
4. [4] A.FUKAO, “Minimum Shuffling Model In The BWR Loading Pattern Optimization Code FINELOAD-3,” Proc. ANS Topical Meeting ANFM-IV, Hilton Head Island, South Carolina, (2009)
5. [5] A. FUKAO, S. YOSHIDA and T. YOSHII, “Actual ABWR Core Design Using Loading Pattern Optimization System FINELOAD-3 Minimum Shuffling Option,” Trans. Am. Nucl. Soc., 104, 902(2011)
6. [6]T. YOSHII, A. FUKAO, Y. KURODA, S. JAGAWA, R. GRUMMER*1 and F. CURCA-TIVIG*2, “Coupling of Boiling Water Reactor Core Loading Pattern Optimization System FINELOAD-3 with MICROBURN-B2,” Trans. Am. Nucl. Soc., 104, 906(2011)
   ^*1: AREVA NP Inc. ^*2: AREVA NP GmbH
7. [7] S. JYAGAWA, ’’Automatic Shuffling for BWR Core Design Using SIMULATE-3,’’ Trans. Am. Nucl. Soc., 62, 514 (1990)
8. [8] RHODES III, J.D., SMITH, K.S., BAHADIR, T., “SIMULATE-3 Speed-Up Models,” Studsvik/SOA-97/10, Studsvik of America, Inc., Newton, MA. (1997)
   http://www.studsvikscandpower.com