Pulsed Neutron Measurement of Single and Two-Phase Liquid Flow

Use of radioactive tracers for flow velocity measurements is well developed and documented. Measurement techniques involving pulsed sources of fast (14 MeV) neutrons for in-situ production of tracers can be considered as extensions of the old methods. Improvements offered by these Pulsed Neutron Activation (PNA) techniques over conventional radioisotope techniques are (1) non-intrusion into the system, (2) easier introduction and better mixing of the tracer, and (3) no requirement to handle large amounts of relatively long lived radioactive materials. Just as in conventional tracer techniques, flow velocity measurements by PNA methods can be based on the transit-time or the total-count method. A very significant difference of the PNA technique from conventional methods is that the induced activity is proportional to the density of the fluid, and that PNA techniques can be used for density measurements (of two-phase flows) in addition to flow velocity measurement. Original equations were derived that relate experimental data to the mass flow velocity and the average density. The accuracy of these equations is not effected by the flow regime. Experimental results are presented for tests performed on liquid sodium loops, on air-water loops, on the EBR-II reactor and on the LOFT reactor. Current instrumentation development programs (detectors, pulsed neutron sources) are discussed.


Introduction
The use of tracers for the study of various aspects or fluid flow goes back to the beginning of this century, when filaments of dye were used to visualize flow patterns in transparent pipes and open channels.Allan and Taylor* were among the first researchers to inject salt into water mains and to determine the flow velocity of the water from the time of arrival of the salt at a detector located downstream of the injection point.Hull and Kent 2 used radioactive tracers to measure flow velocities in oil pipelines.Several good surveys of the various methods used today are available.3.4,5,6,7   Injection of tracers is done either continuously or at discrete times, and flow velocities can be measured from the dilution of the tracer concentration or from the average transit-time of the tracer from the point of injection to the point of detection.The equation used for the transit-time method is Q = -, where T V • Volume (of the vessel or pipe) between the injection plane and the detection plane, and T » Average transit-time (between the injection plane and the detection plane), or average residence time (in the vessel or pipe).
Short lived tracers can be produced in-situ by irradiating the fluid with neutrons.Neutron sources with continuous output as well as pulsed neutron sources can be used.
A large number of studies (including the classical work by Taylor 8 * 9 ), deals with the dispersion of the injected tracers in time, and with numerical methods for the derivation of an "average transit-time" suitable for use J.n Eq. 1.These methods derive an average transit-time T from the time distributed number of counts C(t), by using the mode, the median, or the mean T, where  (3)   It should be pointed and that C(t) in the Eqs. 2 and 3 is the decay corrected number of counts per time channel, which is related to the (background corrected) measured number of counts, c(t), as X = Decay constant and, t = Time after irradiation of the fluid.
An urgent need exists in reactor research for the measurement of the mass flow of sodium, water, and steam-water mixtures.The feasibility of Pulsed Neutron Activation (PNA) techniques for the measurement oi liquid sodium flow was studied at ANL 10 , and it was found that the transit-tine method of the PNA technique was best suited for this application.Me have then applied the transit-time method to the measurement of two-phase flow (air-water, steam-water).The success of our experiment is, basically, attributable to the availability of a very compact and portable neutron source developed by the Sandia Laboratories for a classified program not related to our work.

Equipment
For the measurement of mass flow velocities in pipes, a pulsed neutron source is positioned close to the pipe and the fluid opposite the source is activated by a short burst of neutrons (see Fig. 1).Ideally, the induced activity is originally confined to a narrow slice dZ, and distributed uniformly over the cross section of the pipe.However, the original distribution of activity is changed as the activated fluid moves down the pipe, and a detector located downstream of the pipe will register a distribution of counts as the activated fluid moves past it.Conventionally, the average mass flow velocity is then determined by deriving an average transit time for the activated slug of fluid, according to Eq. 1. Assuming that the cross sectional area of the pipe is constant over the whoJ.e length Z_, both the flowrate of the fluid, Q, and.the volume of the pipe, V, can be expressed by the cross sectional area F, as Q -U.. F, and V « ZQ F, where UH = Mass flow velocity.
Then, Eq. 1 becomes U M " *0 T ' < 5) Sandia developed neutron sources, Model TC655, used in our work.These sources are of cylindrical form, have a diameter of 20 cm and are 35 cm long.They weigh 11 kg.Their neutron output is 3 x 10 9 neutrons/pulse.In experiments of long duration, we pulsed these sources once per minute.Under such conditions, the neutron output remained stable over 400 pulses.We have also pulsed these sources a few times allowing for recovery time of only 15 sec.The only equipment necessary for the operation of these sources is a 28 VDC, 5A power supply, which makes the whole system easily portable.The pulse duration of the source is classified, but the neutron burst can be considered to be.infinitely short when compared with 1 ra sec, the order of magnitude of the window width of multiscalers used in PKA tests.equipment.

Primary PNA Equations
These PNA equations, derived below for two-phase flow conditions but equally valid for single-phase flows, measure independently the average mass flow velocity and the density of fluids.In contrast to conventional two-phase flow measuring system, which measure local velocities and densities, PNA techniques measure "global" velocities and densities and are not affected by specific characteristic of the flow regime or the flow profile.In two-phase flows, velocities and densities can change along the length of test section and can be described as in Fig, 2, where U and Ufi£ are the superficial velocities of the liquid and tne gas (with densities of p attd Pfi)> U..s is the superficial mass flow velocity (the sum of the liquid and the gas flow), U_, pg and U_, p are the velocities and densities at the source location and the detector location, and IL,, p_ are the "transit" velocity and "transit" density, x.e., the velocity and density averaged over the length of the test section between the source and the detector.
The neutrons produced by the pulsed sources used In our measurements are of the D(T,n)a type with an energy of E =14 MeV, and are very effective for a activation of water as well'as sodium.For the activation of water, the reaction 0 16 (n,p) N 16 is utilized, which results in a radioactive nitrogen isotope that decays with a half life of th = 7.2 sec under emission of a gamma ray with an energy of E =6.1 MeV. 11The cross section for the (n,p) reaction is 42 mb, and the threshold is 10.2 MeV.
Several reactions are possible for 14 MeV neutrons with sodium. 11The reaction Na 23 (n,a) F 23 has the highest cross section and the shortest half life of the resulting radioisotope.Therefore, it is this reaction that can be best used for the measurement of liquid sodium flow velocities.The gamma rays produced by this reaction, having energies of E = 1.6 MeV, must be discriminated from the gamma rayS produced by thermal neutron capture in sodium, which have an energy of 2.75 MeV.

Analytic Modeling
In our early measurements, we have used Eq. 3 for the calculation of the average transit-time.This equation, although accurate for conditions where the counts are nearly normally distributed, was found to be inaccurate for application to two-phase flow conditions.In annular flows, for example, the liquid ring close to the pipe wall flows much slower than the mass at the center line of the pipe.In inany situations, there is a distinct separation of the velocities of the two phases and the concept of "slip" can be used.For these complex two-phase flow conditions, a new set of two primary PKA equations was derived that describes accurately the mass flow velocity and the density of two-phase flows in terms measurable by PNA 2. In Eq. 6, the transit-time T is equal to the "time" t, since the time sweep of the multiscaler is started at the time of triggering the neutron source.
Assuming that the neutron fluence over the cross section of the pipe is constant, then the activity a(T) introduced into m(T) by activation is related to the total activity, A, introduced into the total mass, M contained in dZ (see Fig. 1) as (7)   where M_ = The mass contained in dZ when the pipe is waterfilled, and A ™ The activity induced In M_ by neutron activation.
a(T) -£ Now, when the activity a(T) passes the detector window AZ with a velocity U_, (see Figs. 1,2), it will deposit a number of counts C(T) that is proportional to this activity and inversely proportional to its velocity: where C_ = The number of counts registered by the detector when A_ passes the detector window with a velocity of U o = 1 m sec.
Substitution of Eq. 9 for a(T) in Eq. 8 leads to C 0 U 0 and Eq. 6 reads C(T)

XX C(T)
The velocity H_ across the detector window AZ is not known.However, if care is taken to assure that the flow over the whole test length Z Q is constant, then the flow velocities at the detector, the source, and over Z can be assumed to be identical, or U D " U T " V T and Eq.11 becomes (12) (13)   Discussion of the validity of the assumptions made in deriving Eq. 12 as well as detailed demonstrations that Eq. 12 indeed measures the true mass flow velocity, independent of variations of flow regimes, is given in other publications. 12' 13 The second primary PNA equation expresses the density of the fluid in terms of measurable activities.It is derived by equating the total activity injected by the neutron burst to the total activity detected by the detector, as shown below.
In two-phase flow, the density, p, and the mass M, contained in the source window dZ (see Fig. 1) are related through the volume V subtended by the neutron beam as M = V p, and the amount of activity injected into the fluid by a neutron pulse, A,,, can be expressed

On the other hand, the total activity passing the detector, A., is
In past work, it was found convenient to describe the parameters of a particular PNA setup by a constant K, K c o"o (17)   Using Eq. 9 for a(T) and equating the two activities A and A_ leads to This K-value describes the source strength and source collimation,.detector size and detector collimation, discriminator setting, pipe diameter and all other details of an experiment.It is determined by measuring the total number of counts, C , for a waterfilled pipe (p Q = 1) and a flow velocity of V = 1 ml sec.Using the definition for K, Eq. 16 becomes (with Eq. 12) (18)   This equation, then, is a measure for the density of a steam-water mixture in a pipe, available by evaluating the counts/time distribution of a PNA test.

Secondary PNA Equations
Besides of the two primary PNA equations derived above, two secondary equations were found to be very useful for PNA data reduction.One is the well known equation relating the total number of counts to the flow velocity, which is simply derived by substituting general values C = I C(T), U, and p, for the specific values C in Eq. 17. ( The physical meaning of the average velocity U can be understood by deriving Eq. 19 from Eq. 5, with a definition for the average transit-time as in Eq. 3. A second, independent determination of the density can be obtained by PNA methods to verify the measurement done by Eq. 18, whenever measurement of the superficial gas and liquid flow velocities are available.From these two superficial velocities, a superficial mass flow velocity U can be calculated 13 (see Fig. 2), and the density of the fluid p_, averaged over the length Z Q of the test section, is thencalculated from the measured transit-time velocity U (Eq. 13) by (20)

Experimental Data
Measurements .-it the EBR-II Power calculations performed from data taken on the primary and the secondary loops of the Experimental Breeder Reactor-II (EBR-II) in Idaho Falls have shown a 9% discrepancy. 11* To resolve this discrepancy, PNA measurements were performed on the secondary sodium loop of the reactor, which has a diameter of 30 cm.Our experiments have shown 15 ' 16 that, at full flow, the actual secondary flow is 1.057 times higher than the flow indicated by the permanently installed flow instrumentation.The discrepancy existing between primary and secondary loop power levels is now only 4%, and at full power operating conditions, the plant power is now calculated at 60.1 MWt.
The water flowrate in the EBR-II evaporatordowncomer was also measured by PNA techniques. 16' 17 Although the PNA measurement at the EBR-II were made in a relatively high radiation background (due to the activation of sodium by thermal neutrons), the accuracy of the flow measurements was still about 1%.The agreement between predicted counting rates and measured counting rates was good. 15asurements at the ETEC The performance of the large sodium puraps of the Fast Flux Test Facility (FFTF) is being evaluated at the Energy Technology Engineering Center (ETEC) at Canoga Park, California.The permanent magnet flowmeters installed on the 40 cm pipes of the Sodium Pump Test Facility (SPTF) at the ETEC is routinely calibrated by readings from a venturi flowmeter installed in the same line.The venturi flowmeter, in turn, is calibrated in a water loop with known flowrates.Since removal and re-installation of the venturi meter is cumbersome and costly, PNA flow measurements were made and compared with venturi meter readings.Test data (Fig. 3) have shown that PNA data are as accurate as the venturi data. 16It was decided, therefore, to install a PNA measuring system at the SPTF for routine calibration of the permanent magnet flowmeter.Our first two-phase flow measurements were performed at the air-water loop located at the Semiscale facility at the Idaho National Engineering Laboratory (INEL) in Idaho Falls.Tests were performed at two pipe diameters, 7.5 cm and 12.5 cm, and for two sourcedetector spacings.Evaluation of test data has shown 13 that, for the short spacings, there is good agreement between the transit-time velocity and the total-count velocity (see Fig. 4), as well as between the transittime density and the total-activity density (see Fig. 5).Data derived by Eqs. 13, 18, 19 and 20 were used in these two figures.For the long spacings, the agreement between the two sets of velocity and density measurements were not good, due to the change of pressure along the test section.This pressure drop (due primarily to the instrument spool piece installed between the two detectors as well as to other causes) violates the PNA condition shown in Eq. 12.In its present configuration, the LOFT facility is capable of fulfilling a number of test objectives, including the study of Emergency Core Cooling Systems (ECCS's) used as safeguards in events of coolant loss, to remove heat from the core and to prevent damage to the fuel.PNA techniques were used to measure the rel.-.ve ECC bypass flow at the LOFT reactor. 19Four simii'.^ancouslytriggered neutron sources were used in 'chis test and pairs of 12.5 cm diameter x 7.5 cm Kal detectors were used for the registration of counts..Tija experimental setup of the PNA equipment is illustrate** in Fig. 6.The bypass ratio was found to be (41 + 7)2.

tCC INJECTION
Fig. 6 PNA Setup at the LOFT Reactor

Development of Neutron Source
A high-output neutron source is being developed by the Sandia Laboratories for future tvro-phase flow measurement by PNA techniques.Sandia has an established capability not only in the development and construction of classified neutron sources, but also of an unclassified neutron generator for use in uranium logging. 2 " This unclassified logging generator uses a modification of the Controllatron, a General Electric -developed Penning ion source. 21Another type of ion source considered for incorporation in a two-phase measuring neutron source is a solid-state spark ion source originally developed at Berkeley. 22t the present rime, Sandia has three different neutron source configurations under consideration.All three are likely to lead to a pulsed neutron source that will meet the requirements for our two-phase flow measurement program.The first approach utilizes the Berkeley-type, solid-state spark ion source, and a high voltage supply consisting of a transformer assembly and a pulse forming network (PFN).The second configuration under study is a Penning-type gas discharge ion source, again in Combination with a PFN and a transformer assembly.The third neutron source under study uses a Penning-type ion source with a dc high voltage supply (a Cockroft-Walton generator).Pulsing of this neutron generator is achieved by pulsing of the ion source.Within the next few months, one of these three configurations will be selected for production.A prototype of the new pulsed neutron source will be available in mid-1980.The specifications for this unclassified source are: output: 10 10 neutrons/pulse, pulse length: 1 m sec, recovery time: 5 see, lifetime: 1000 pulses.

Development of Torus Detector
A torus detector, fully encircling a pipe, is being developed at ANL and will be available early in 1979.This detector will encircle pipes with diameters of 15 cm, will be 10 cm wide and will have a height of 8 cm.Four photomuleiplier tables will be used to detect the scintillations in the four-segmented Nal Bcintillator.

Fig. 2
Fig.2PNA Nomenclature Now, the first of the two primary PMA equations relates the measured count/time distribution (such as illustrated in the multiscaler display at Fig.1), to the true mass flow velocity of liquids In pipes, averaged over the volume of the pipe between the source and the detector locations.This equation is directly traceable to the definition of the average mass flow velocity, which is

Fig. 5
Fig. 5 Comparison of Local Density, p, to Transit-Density PT