| OF T. ORNLP .2979 . .. . IP . FEZFEFFE 1 11.25 1.1.4 LLS MICROCOPY RESOLUTION TEST CHART NATIONAL BUREAU OF STANDARDS - 1963 ORNU V 29:19 - Conf-670314... C!? DICOS SUBCRITICALITY MEASUREMENTS BY NEUTRON NOISE ANALYSIS L MN APR 28 567 D. P. Roux Oak Ridge National laboratory ELLASED FOR Anoua Oak Ridge, Tennessee | 10 NUCLEAR SCIENCE ABSIRACTS 1. INTRODUCTION 'The use of reactor fluctuations (noise) to measure reactor parameters 16 : well established. The method has been applied to measurement of reactivity of subcritical systems by many workers.2,3 This particular technique is tased on the fact that the frequency distribution and the amplitude of the reactor neutron fluctuation spectrum are functions of the reactivity p, and other reactor parameters. The intent of this paper is to evaluate the applicabllity of this technique for Liquid Metal Fast Breeder Reactors (IMFBR) as an on-line safety device. 2. ACCURACY OF SUBCRITICALITY MEASUREMENTS BY NOISE ANALYSIS In practice, to estimate the subcriticality, the fluctuaticas or the out- put current of a neutron-sensitive ionization chamber are analyzed, and the observed frequency spectrum 18 fitted to a reactor model (the point reactor model is generally used). Excluding analysis equipment error, the resulting spectrum data points contain an uncertainty due to the statistical nature of the noise. The fractional error of each spectrum data point is 1 VI/ET for the power spectral density in a single channel, where B is the bandwidth of the channel filter and T is the analysis time. The selection of the band- width of the filter 18 dictated by the frequency distribution of the reactor *Research sponsored by the U. S. Atomic Energy Commission undez' contract with the Union Carbide Corporation. . . Dit noise spectrum and its breakfrequency. For a critical reactor the break- frequency fy 18 . (2) * where 3. is the effective fraction of delayed neutrons and A 18 the prompt neutron generation time. Typically for thermal reactors fy 16 equal to 20 cps, and for fast reactors fy 1s equal to 2 kc. In addition to the statistical uncertainty effect, the accuracy of the subcriticality measurement is further degraded by a frequency-independent * * * * ututäitminina * background caused by the randomness of the chamber detection process. For a Lii L ediante a kit critical zero-power reactor, at frequencies fa lower thau fy where the reactor spectrum is essentially flat, the ratio of reactor correlated (desired) noise C to the unwanted detection noise U is approximately r m - - L om noise U is approximately **. *. - n * , i KV K (Taup = '0) = max = 0) – 0.5 wale (3) -,- ... ... . .**... ... where Wo is the neutron detection efficiency of the chamber expressed in * neutron detected per fission in the reactor. -- n In a thermal swimming-pool type reactor, in which high detection efli- .. i V o ciencies can be experienced if a sensitive detector is placed at the edge of the core, Qaylo = 0) is equal to ~ 30. For a subcritical reactor, Eq. (3) OY becomes The term (1 - p)2 explains why large subcriticality values (p in dollars) are so difficult to measure with acceptable accuracy. As an example, for analysis times T = 15 min, in a swimming-pool reactor, p can be estimated to #3 cents at -dollar and 30 cents at -5 dollars. LEGAL NOTICE This report m. prepared as an account of Government sponsored work. Neither the United States, aor the Commission, aor nay pornod acting on behalf of the Commission: A. Makes any warranty or representation, expressed or implied, with respect to the accu- racy, completeness, or unofulnou of the information contained in the report, or that the un of any information, apparatus, method, or procon disclosed to the report may not latring privately owned righta; or B. Asamor any liabilities with respect to the use of, or for damage rating trou the use of any information, apparatus, method, or proche declound in the reports As und in the above, "person acting on behalf of the Caninaloa" tncludes My On- ployer ur contractor of the Commission, or soaployee of mch contractor, to the adent that such employee or cotractor of the Commission, or employee of much coatructor prepare, disseminates, or provides access to, kay Information purouant to Mo saploymeat or contract with the Commission, or his employmont with such contractor. 3. RECENT WORK AND STUDD'S AT OAK RIDGE It is essential to consider the potential value of this technique for on-line measurements in power reactors. In this application three factors can severely reduce the accuracy with whick. p can be estimated: 1. fission product gamma backbround, which produces an additional white noise component; 2. 103 detection efficiency due to the large physical size of power reactors, and problems encountered with detector location close to the core; 3. short sampling times for on-iine application (see Eq. 1). 3.1 Gama Background To alleviate the problem due to gamma background, we have investigated three approaches : 1. A puise detector was used rather than an ionization chamber to dis- criminate against gammas. The current fluctuations are produced in a CR circuit after the discriminator. We demonstrated experimentally the equivalence of the two methods (pulse and current modes) by putting a counter and an ioniza- tion chamber side by side and measuring the subcriticality of a clean zero-power reactor for various subcritical conditions (up to approximately -$4.5). Then, using a large cobalt source, we simulated the effect of the fission product gamma background ; and with boron-lined proportional counters, we measured subcriticality in a gamma flux up to 105 r/hr. Also, we investigated the problem of counting loss, and demonstrated that counting losses up to 20% are still acceptable. 2. In the second approach, we made an analys18to quantitatively evaluate the gamma degradation in the current mode detection. The results were as follows: 0.5 .-2 @max(s) ~ TI -)[1 4 57:32] (5) Equation (5) is identical to Eq. (4) except for the term [1 + (0 )2] which is the ganuna degradation actor. 2 is a detector characteristic. We rade some studies to minimize z and determined that Z can be as small as 2 in flux units for thermal reactors. Thus, as long as the gamma flux (ir r/hr) is not very much larger than the neutron flux (in nv), the gamma degrada- tion is tolerable. 3. In the third approach we applied the two-detector cross-correlation method (2DCM). 6 When the outputs of the two detectors are multiplied, the uncorrelated noise portion of each detector signal, whether originating from the basic detection procese6 or from the gamma rays”, is cancelled out on the . average. Unfortunately this process of cancellation requires a greater amount of analysis time the lower Qemay becomes, and it accordingly does not seem to provide the advertised "breakthrough" for on-line application, where analysis time is limited to a few minutes by practical considerations. In this case Qmoulo) must be * 0.5 to obtain acceptable subcriticality estim tes. 3.2 Estimate of Statistical Precision of Subcriticality. Measurements Until recently we could not reliably predict the outcome of a subcritical measurement experiment. Now that we have developed an analytical procedure, 8 we can predict the statistical precision of a subcriticality measurement for a given set of conditions: Woo Bee Ano No 2, 1, and available sampling time. In other words, if these parameters are fairly well known, we can predict the statistical precision of the subcriticality measurement in terms of so many cents at so much dollars subcritical. 4. SUBCR.MICALITY MEASURW.EITS IN LMFBR Although the following discussion is pertinent to the LITBR Jiropram, only speculations can be made. However, some specific problems to be enticipated with fast large reactors (1000 MW) of this type are examined l the following discussion. 4.1 Detection Efficiens:y For reestors of this type, the situation regarding detection efficiency 1s summarized as follows: 1. It is a little grim in general, because of the relative geometric size of the detector compared to the large size of these systems. 2. The lowered detection cross section for fast neutrons compared to thermal neutrons will be somewhat compensated for two reasons. Firist, A lower By In fast reactors will increase Qmax (Eq. 3) by roughly & factor of 5. Second, the breakfrequency is situated in the range from 1 to 10 kc rather than at 20 cps as discussed in Sect. 2; thus, filter bandwidths 100 times max larger can be used and this an additional gain of a factor of 10 in statistical precision is expected (see Eq. 1). 4.2 Equipment Problems Related to Higher Spectrum Breakfrequencies There are advantages at higher spectrum breakfi'equencies in that there is no need for very Low-noise amplifiers, since amplifier noise is proportional to p-; and the 60-cps stray pickup problem, which is severe in thermal reactors, will be eliminated. However, there are also difficulties at higher spectrum breakfrequencies. The frequency response of the equipment in the range from 1 to 50 kc presents two problems: (1) since lon mobility in the lonization chamber causes rolloff usually at ~1 kc, special chambers must be developed or pulse mode must be used; and (2) 1f the current mode 16 used, and to avoid the chamber RC rollofl, current input amplifiers will be required rather than voltage amplifiers. Another source of difficulty inherent in IMFBR measurements is that the detector will be exposed to a high temperature environment. 4.3 Validity of Point Reactor Model Approximation Both the multiplication and neutron noi.se methods used to predict subcriticality are based on the point reactor approximation. This validity 18 questionable for large LMFBR, and studies should be made. 5. CONCLUSIONS Reactor subcriticality measurement by noise analysis has a fair chance to be successfully applied in LMFBR's if the measurements are limited to only . . S ..* * 25 small shutdown margin measurements. In the range from 1 to 2 dollars sub- critical, an accuracy of 10% can be expected. However, as pointed 'out in Sect. 4.2, a few severe detection problems will have to be solved. . A - - : . . . - - - - - - R E M REFERENCES 1. J. A. Thie, Reactor Noise, Rowman and Littlefield, New York, 1963. 2. R. E. Uhrig, ed., Proceedings of the Symposium on Noise Analysis in Nuclear Syetems, AEC Symposium Series 4 (1965). 3. C. W. Ricker et al., Measurement of Renctor Fluctuation Spectra and Subcritical Reactivity, ORAL-TM-1066 (April 1965). Le J. R. Irinko et al., "Reactor Noise Using a Pulse Type Detector," Trans. Am. Nucli. Soc. 10 (1), to be published. 5. D. P. Roux, "Optimization of Reactor Shutdown Margin Measurements in High Gamma Fluxes," Trans. Am. Nucl. Soc. 9(2), 523 (1966). 6. W. Seifritz et al., "TWO-Detector Cross Correlation Experiments in the Fast-Thermal Argonaut Reactor Stark;" paper to be published in the Proc. Intern. Symp. Neutron Noise, Gainesville, Florida, Feb, 14-16, 1966. 7. R. C. Kryter et al., "TWO-Detector Cross Correlations for Shutdown Margin Measurements in Ganma Fluxes," Trans. Am. Nucl. Soc. 2001), to be published. 8. D. P. Roux and R. C. Kryter, "Two-Detector Cross Correlation for On-Line Shutdown Margin Measurements in Power Reactors," Trans. Ame Nucle Soce 10 (1), to be published. - -- - END DATE FILMED 5 / 26 / 67 SLAI ... . Ft, . ELS." ist doen het kommer manier stränder mer . .ne .. ..