Spectroscopic Based Process Monitor for Contin o s Real Time Anal sis of HighContinuous Real-Time Analysis of High Level Radioactive Waste Components and Monitoring for Control and Safeguarding o to g o Co t o a d Sa egua d gof Radiochemical Streams

Sam BryanSam Bryan

Eastern Oregon UniversityLa Grande, OregonJanuary 29, 2010

1

Outline

Background of Hanford SitegProcess Monitor for Continuous Real-Time Analysis of High Level Radioactive Waste ComponentsMonitoring for Control and Safeguarding ofMonitoring for Control and Safeguarding of Radiochemical Streams

2

Hanford Site relative to San Francisco Bay Area(same scale)(same scale)

3

Hanford Double Shell Tank

• LiquidSalt cake• Salt cake

4

Waste Retrieval from Nuclear Waste Storage Tanks

Composition Storage andwater spray fordissolving salt cake

pMonitoring

gVitrificationdissolving salt cake

Salt cake

5

Salt cake

Photograph of Interior of Single Shell Hanford TankHanford Tank

6

Tank Farm Chemical Inventory, Hanford SiteTank Farm Chemical Inventory

Na

NO3

OHOHNO2

CO3

by phase

Al saltsAl

PO4

SO4 NaNO

NaNO2

Na-CO -F-Fe

TOC

F

NaNO3-PO4-SO4

Na-CO3-F-

0 10000 20000 30000 40000 50000 60000

others others

7

Metric TonsBBI 3/20/03

Current Retrieval Capability

FlowGamma monitorGamma monitorGrab samples

OnOn--line Realline Real--time Process Monitortime Process MonitorOnOn--line Realline Real--time Process Monitortime Process MonitorContinuous real-time monitoring of retrieval composition Immediate status and composition of retrieval Retrieval summary from start of processing

Process monitor benefits process performancep p cost savings doesn’t require personnel involvement to grab samples doesn’t interrupt continuous retrieval

8

doesn t interrupt continuous retrieval easily adaptable for variable waste applications

Process Monitor Features

Instrumentation was configured for remote monitoring (spoolpiece, fiber optic, cables)

Coriolis and conductivity metersCoriolis and conductivity metersStraight data readout; density, flow, conductivity

RamanOutput spectral files

Methodology of data processingChemometric data treatmentTwo independent chemometric models

Total sodium concentration (conductivity, density, temp)8 individual sodium salt concentrations (Raman spectra)

NO3-, NO2

-, PO43-, CO3

2-, CrO42-, OH-, SO4

2-, Al(OH)4-

Data acquisitionData storage-archivingDisplay

9

MatLab-based software to process, store, and display data

Measurements needed for the development of chemometric modelof chemometric model

Raman signal conductivity and density depend upon:Raman signal, conductivity, and density depend upon:Solution composition and ionic strength

Measure solutions of individual Na salts and mixtures in wide concentration rangeconcentration range

TemperaturePresence of bubbles and solid particulates

Instrument performance at variable flow rateMeasure solutions in static mode and under variable flow rate

Acceptance Test: Verify chemometric model using early and late feed simulant

10

Verify chemometric model using early and late feed simulant solutions

Static Measurements

86 solutions were prepared and measured by Raman, Coriolis, and conductivity sensors at room and several elevated temperatures.Solids as Fe(OH) or bentonite were added to the selected solutions andSolids as Fe(OH)3 or bentonite were added to the selected solutions and measured.Preliminary chemometric model was developed using static measurements.Single component and multiple mixture standards usedSingle component and multiple mixture standards used

component concentration rangeNO3

- 4 – 40 wt%NO3 4 – 40 wt% NO2

-, 3 – 34 wt% PO4

3-, 1.5 – 15 wt%CO3

2-, 1.5 – 15 wt%3 ,CrO4

2-, 0.8 – 8 wt%OH-, 1 – 20 wt%SO4

2-, 2 – 20 wt%

11

Al(OH)4- 0.8 – 8 wt%

Flow Mesurements

Total 6 flow tests were performed:5 tests used laboratory bench top flow loop;5 tests used laboratory bench-top flow loop;1 test combined field manifold and laboratory flow loop.

Flow rate studied; 2 - 20 GPM range.Temperature was varied from 15 to 50°C.N2 Bubbles were introduced into flow system at controlled variable rates up to 6% of the solution flow rate.Eff t f F (OH) lid t di d t 1% lid l diEffect of Fe(OH)3 solids was studied; up to 1% solids loading.Revised Raman chemometric model incorporated flow measurements.Sodium model for prediction of total Na concentration was developed b d d ti it / d it / t t tbased on conductivity / density / temperature measurements.

12

Photo of Process Monitor EquipmentPhoto of Process Monitor Equipment

Coriolis

conductivitymeter

Ramanprobe

Co o smeter

flow inlet to pump

flow outlet from system

13

y

Raman Spectra of Ionic SpeciesRaman Spectra of Ionic SpeciesRaman Spectra of Ionic SpeciesRaman Spectra of Ionic Speciesp pp pp pp p

NO3NO3 CrO4

SO4

NO2CO3H2PO4

400 600 800 1000 1200 1400 1600 1800 2000Raman Shift (cm-1)

14

Temperature influence on spectra

variable temperature nitrate solutionsnitrate region is insensitive to temperature variation

200030004000500060007000

man

inte

nsity

010002000

1000 1020 1040 1060 1080 1100

wavenumbers, cm-1

Ram

variable temperature nitrate solutionspwater region is sensitive to temperature variation

150020002500

ensi

ty

22°C

0500

10001500

3000 3200 3400 3600 3800 4000

Ram

an in

te

58°C

15

wavenumbers, cm-1

Added Solids Reduce Raman Signal

Mix 12 with added Fe(OH)3 solids Mix 12 with added Fe(OH)3 solids

3000

35004000

45005000

nsity

, cts

Series1Series2Series3Series4Series5Series6

Aluminate

1000

1200

1400

1600

nten

sity

nitrate region

NO3-

0500

10001500

20002500

Ram

an in

ten Series7

Series8Series9Series10Series11Series12Series13Series14Series15

NO2-

water

0

200

400

600

800

Ram

an in nitrite region

water regionAl(OH)4-

0500 1500 2500 3500

wavenumber, cm-1

Series16Series17

0 100 200 300 400 500

uL Fe(OH)3 added

16

Raman Model

P i I Cl i l L t S b d d lPrimary Ions use Classical Least Squares-based model Calibration steps

Normalize to OH stretch (to correct for intensity variations) then use 1st derivative (to reduce impact of luminescence backgrounds)Use only low spectral range (salts 517-1836 cm-1; OH- >2670 cm-1)Estimate "pure component spectra" (what each species would look like if it were alone in solution) for each ionic speciesAllow more than one possible spectrum from each species (becauseAllow more than one possible spectrum from each species (because some spectral peaks shift in highly concentrated solutions). Allows approximation and handling of "non-linear" spectral behaviorInclude additional "background" spectra (estimates of interferences) to account for other backgrounds

0.2

0.25

Normalization reference,then use for OH- ion only

Take 1st derivative anduse for all other ions.

0.05

0.1

0.15

17

1000 1500 2000 2500 3000 3500 40000

Raman Model Prediction of ConcentrationsRaman Model Prediction of ConcentrationsRaman Model Prediction of ConcentrationsRaman Model Prediction of Concentrations

60tion

40

50

60

Con

cent

rat

NO3-1

mix-1 mix-5

20

30

Tota

l Ion

C

NO2-1

SiO4-2

mix-3

mix-7

mix-12

mix-14

mix-16

mix-22

mix-23

i 24mix-35

mix-36

mix-37

mix-39 mix-41

mix-43

0 5 10 15 20 25 30 35-10

0

10

Pre

dict

ed T

AlO2-3 PO4-3 NaOH-1 RNaOH-6 R

mix 22 mix-24 mix-26

mix-30 mix-33 mix-34

mix-40

0 5 10 15 20 25 30 35

Actual total Ion Concentration

P

total ion concentration displayed (sum of all calculated species)

18

total ion concentration displayed (sum of all calculated species)

Sodium Model

Measure: Density (x1), Conductivity (x2), Temperature (x3)Create non-linear combination of these variables to the data (x1

2, 2 2 / / / / ) Gi t t l f 15x22, x3

2, x1x2, x1x3, … , x1/x2, x1/x3 , x2/x1, x2/x3, ...) Gives a total of 15 variables.Use Genetic Algorithm to select "best" variables (those that give lowest error of cross-validation from PLS model))Results include 5 variables: density, density2, density*conductivity,

conductivity*temperature, density/conductivityUse Partial Least Squares (PLS) and Locally-Weighted R i (LWR) t di t di t tiRegression (LWR) to predict sodium concentration.LWR builds "local" regression model from calibration data which looks

most like the unknown.Chemometric model developed by Eigenvector Research IncChemometric model developed by Eigenvector Research, Inc.

19

Locally Weighted Regression Models

1. Given PCA model of all calibration data and an unknown sample, select the 60 calibration samples closest to the unknown (green

4

5

closest to the unknown (green samples).

2. Use ONLY these samples to build Principal Component Regression model (from whole-data scores)

1

2

3

PC

1 (7

3.17

%)

model (from whole data scores) and predict.

3. Refine sample selection based on estimated Y and known Y from calibration samples

-2

-1

0Sc

ores

on

P

calibration samples.

-5 -4 -3 -2 -1 0 1 2 3-4

-3

Scores on PC 2 (23.73%)

20

Calibration Results PLS vs. LWR

8

9

ons

8

9

327

3422 Latent Variables (LV)RMSEC = 0 37

Calibration Data vs TimeCalibration PlotsPLS

4

5

6

7

ross

-val

idat

ed P

redi

cti

4

5

6

7

date

d P

redi

ctio

ns

132

139

327

349

353356

360

446RMSEC = 0.37RMSECV = 0.38 ~ 0.38M error

Variable

100 200 300 400-1

0

1

2

3

Mea

sure

d N

a, C

r0 2 4 6 8

-1

0

1

2

3

Cro

ss-v

ali

10

47

7988

360364 •composition

•temperature•gas additionfl100 200 300 400

1

Sample0 2 4 6 8

1

Y Measured 1 Na

6

7

8

9

ted

Pre

dict

ions

6

7

8

9

dict

ions

RMSEC = 0.04RMSECV = 0.06

2 LVs & 60 Samples

•flow rate•solids addition

Data (blue)

flow test 1

2

3

4

5

6

ured

Na,

Cro

ss-V

alid

at

2

3

4

5

6

Cro

ss-v

alid

ated

Pre

d Data (blue)Fit (green)

LWRflow test 2

21100 200 300 400

0

1

Sample

Mea

su

0 2 4 6 8 100

1

Measured

~ 0.06M errorflow test 2

Compositions of S109 Feed Simulants Used in Acceptance Tests

Early Feed Late Feed

Analyte M %Wt M %Wt

Density 1.42 1.15 OH 1.67 4.70 as NaOH 0.021 0.073 as NaOH

TOC <0.03 <0.03 TIC 0.61 4.55 as Na2CO3 0.24 2.21 as Na2CO3

Al 0.52 4.32 as NaAl(OH)4 0.044 0.45 as

NaAl(OH)4 Cr 0 079 0 92 as Na2CrO4 0 018 0 26 as Na2CrO4Cr 0.079 0.92 as Na2CrO4 0.018 0.26 as Na2CrO4Fe 0.0001 <0.00002 K 0.026 0.009 Na 7.87 12.75 2.62 5.24 P 0.025 0.052 S 0.13 0.17 Si 0.0005 <0.0002 F <0.016 0.098

Acetate NA NA Cl 0.08 0.33 as NaCl 0.009 0.051 as NaCl

NO2 0.86 4.18 as NaNO2 0.073 0.44 as NaNO2 NO3 3.98 23.82 as NaNO3 1.59 11.75 as NaNO3 PO4 0.025 0.29 as Na3PO4 0.055 0.78 as Na3PO4 SO4 0 14 1 40 as Na2SO4 0 17 2 10 as Na2SO4

22

SO4 0.14 1.40 as Na2SO4 0.17 2.10 as Na2SO4Oxalate <0.031 0.011

Cs (µg/mL) 2.3 0.23

Chemometric Prediction of Component Concentrations for S109 Feed SimulantsChemometric Prediction of Component Concentrations for S109 Feed Simulants

Saltpred %Wt Prediction

Analyte %Wt

Analytical Average Absolute

Range

RelativeError

%

Absolute Deviation

%Wt

Detection Limit

%Wt (M) Acceptance Test Acceptance Test Criteria:Criteria:20% l ti20% l tiA) Early Feed, Total of 2028 Measurements

NaOH 4.70 4.85 (0.25) 3.8 – 5.9 3.2 0.15 0.38 (0.09) Na2CO3 4.55 5.02 (0.13) 4.4 – 5.4 10 0.47 0.093 (0.009)

NaAl(OH)4 4.32 4.72 (0.11) 4.4 – 5.1 9.3 0.40 0.30 (0.025)

N C O 0 92 0 96 (0 02) 0 92 0 99 4 3 0 04 0 021 (0 001)

20% relative error or 20% relative error or 0.5% absolute error0.5% absolute error

Two solutions of Two solutions of t itit itiNa2CrO4 0.92 0.96 (0.02) 0.92 – 0.99 4.3 0.04 0.021 (0.001)

NaNO2 4.18 4.22 (0.04) 4.1 – 4.3 1.0 0.04 0.19 (0.03)

NaNO3 23.82 23.47 (0.17) 23.0 – 24.0 1.5 - 0.35 0.033 (0.004)

Na3PO4 0.29 0.35 (0.04) 0.21 – 0.64 21 0.06 0.29 (0.02)

N SO 1 40 1 43 (0 01) 1 4 1 5 2 1 0 03 0 047 (0 003)

extreme compositions extreme compositions were testedwere tested

Test included real time Test included real time l ti f 2800l ti f 2800Na2SO4 1.40 1.43 (0.01) 1.4 – 1.5 2.1 0.03 0.047 (0.003)

Na 12.75 12.87 (0.04) 12.8 – 13.0 1.0 0.12

B) Late Feed, Total of 779 Measurements

NaOH 0.073 0.0 (0.29) 0.0 – 1.1 - 0.073 0.38 (0.09) Na2CO3 2.21 2.31 (0.04) 2.2 – 2.5 4.5 0.10 0.093 (0.009)

evaluation of ~2800 evaluation of ~2800 independent independent measurements measurements collected over 24 h time collected over 24 h time i t li t l3

NaAl(OH)4 0.45 0.41 (0.03) 0.28 – 0.52 8.9 - 0.04 0.30 (0.025)

Na2CrO4 0.26 0.19 (0.01) 0.17 – 0.29 27 - 0.07 0.021 (0.001)

NaNO2 0.44 0.33 (0.02) 0.23 – 0.41 25 - 0.11 0.19 (0.03)

NaNO3 11.75 11.89 (0.10) 11.4 – 12.3 1.2 0.14 0.033 (0.004)

intervalinterval

Acceptance test criteria Acceptance test criteria metmet

23

3Na3PO4 0.78 0.75 (0.05) 0.20 – 1.2 3.8 - 0.03 0.29 (0.02)

Na2SO4 2.10 2.0 (0.02) 2.0 – 2.3 4.8 - 0.10 0.047 (0.003)

Na 5.24 5.38 (0.10) 3.9 – 5.9 2.7 0.14

Field Manifold Verification Test

Compare field manifold system with laboratory systemAssemble combined flow systemAssemble combined flow systemConnect combined flow system with pump and 55 gallon feeding/receiving reservoir Connect laboratory instruments and field instruments to i d d t d t i iti t d tindependent data acquisition systems and computers

Verify of the electronic componentsCompare instrument performance

using 5 wt% NaNO3 solutionusing 5 wt% NaNO3 solutioncontinuously changing the composition of the test solution

Test chemometric models using data independently collected by laboratory and field systemslaboratory and field systemsTest software for the data acquisition, processing, storage, and archival

24

Assembled Process monitor manifoldAssembled Process monitor manifold

Hydro-Cyclone

Filter

Valve to Shielded

Tank System

Coriolis Raman

Valve to 241-SY-101

Coriolis Flow Meter

Raman Probe Valve for

Raw Water Supply

C

25

Conductivity Probe

Isolation Valves

Field Manifold Verification Test

26

Field Manifold Verification Test Results

Field unit calibrated using previously calibrated laboratory itunit

Field system instrumental performance was comparable t l b t tto laboratory system

Composition change of test solutions was successfully p g ymonitored by both systems

Software performed equally well for field and laboratorySoftware performed equally well for field and laboratory systems

27

Schematic presentation of the software Schematic presentation of the software computer displaycomputer display

Retrieval SummaryCurrent ValuesCurrent Retrieval Composition,

computer displaycomputer display

yTime, (days)Volume, in gallons and litersTotal Mass,in pounds and KgSodium in pounds and Kg

Time of last measurementTemp, F and CFlow rate, GPMDensity SGU3.0

3.54.04.55.0

entr

atio

n (%

)

p ,As Sodium Salts

Sodium,in pounds and KgNaNO3 NaNO2Na2SO4 Na2CO3Na2CrO4 Na

Density, SGUConductivity, mSCurrent wt% Na0.0

0.51.01.52.02.5

Con

ce

Na3PO4 NaOH24-Hour Summary of RetrievalMajor Components

G d l %Given as trend plot wt% vs time

24-Hour Summary of RetrievalMinor Components

28Given as trend plot wt% vs time

On-Line Monitoring for Control and Safeguarding of Radiochemical Streams at S F l R i PlSpent Fuel Reprocessing PlantApplication of past on-line process monitoring experience to UREX flowsheet

Spectroscopic measurements of selected fuels and fuel simulants

Evaluation of detection limits

Initial development of quantitative PLS models for U Pu and NpInitial development of quantitative PLS models for U, Pu, and Np

Demonstration of process monitoring using commercial fuel

Instrumentation of a centrifugal contactor system with Raman and vis-NIR spectroscopic probes

Approach: Online Spectroscopic MeasurementsOnline Spectroscopic Measurements

Raman measurements ofA ti id id iActinide oxide ions Organics: solvent components and complexantsInorganic oxo-anions (NO3

-, CO32-, OH-, SO4

2-, etc)Water acid (H+) base (OH-)Water, acid (H+), base (OH )

UV-vis-NIR measurements oftrivalent and tetravalent actinide and lanthanide ionstrivalent and tetravalent actinide and lanthanide ions

Potential UsesProcess monitoring for safeguards verification (IAEA)Process monitoring for safeguards verification (IAEA)Process control (operator)

Previous Experience: Hanford SitepReal-time, online monitoring of high-level nuclear wasteRaman spectroscopy, flow, temperature, conductivity, density

Process Monitoring Can Be Achieved Through-out the Flowsheet

dissolved fuel

Global vision:UREX U

TcProcess monitoring/control at various points in flowsheet is feasible

CCD-PEG

NPEX

Cs/Sr

Np/Pu

Every flowsheet contains Raman and/or UV-vis-NIR active species

TRUEX FPs

active species

Coriolis and conductivity instruments can be used on

Cyanex 301 Am/Cm

instruments can be used on all process streams

rare earths Monitoring Is Not Flowsheet Specific

Methodology for on-line process monitor development: from proof of concept to final outputfrom proof-of-concept to final output

Static measurements:Model training database

Chemometric model development

On-line model verification and

0.4

0.5

0.6

0.7

orba

nce

Pu chemometric modelPLS model for Pu(IV) using UV-vis data

y = 0.9997x + 0.0006R2 = 0.9999

8

10

12

pred

icte

dM

, mod

el

Model training database development verification and translation

0

0.1

0.2

0.3

450 550 650 750 850 950

wavelength nm

Abs

o

0

2

4

6

0 2 4 6 8 10 12

Pu(IV), mM

Pu(

IV),

mM

[P

u(IV

)], m

M

wavelength, nm ( ),

Real-time on-line concentration data display

Integrated software for data collection, processing, storage and archiving

Raman Spectroscopy for Monitoring Uranyl and Nitrate ionsRaman spectroscopy allows for quantification of UO2

2+ and NO3- in aqueous

and organic phases20000

UO22+

Comparison of loaded TBP/dodecane and aqueous feed

and organic phases

10000

15000

n in

tens

ity, a

u

NO3-

dodecaneTBP

UO2(NO3)2 in 0.8 M HNO3

100000

Variable UO2(NO3)2 in 0.8 M HNO3UO2(NO3)2 in 0.8 M HNO3

100000

Variable UO2(NO3)2 in 0.8 M HNO3

0

5000

750 1000 1250 1500

Ram

a

60000

70000

80000

90000

100000

pons

e

1.310.660.49

0 16

UO22+

NO3- UO2(NO3)2, M

spon

se

60000

70000

80000

90000

100000

pons

e

1.310.660.49

0 16

UO22+

NO3- UO2(NO3)2, M

spon

se

20000

UO22+ before extraction

750 1000 1250 1500

wavenumber, cm-1

20000

30000

40000

50000

Ram

an re

sp

0.160.0330.016

0.00330.0016

0.0003

Ram

an re

s

20000

30000

40000

50000

Ram

an re

sp

0.160.0330.016

0.00330.0016

0.0003

Ram

an re

s

Quantitative extraction assessment

5000

10000

15000

Ram

an in

tens

ity, a

u

UO22+ after extraction 0

10000

800 850 900 950 1000 1050 1100 1150

wavenumber, cm-1

wavenumber, cm-1

0

10000

800 850 900 950 1000 1050 1100 1150

wavenumber, cm-1

wavenumber, cm-1

0

5000

825 850 875 900 925

wavenumber, cm-1

Detection limits: 3.1 mM for UO2 2.6 mM for NO3

Vis-NIR spectra of variable Pu(IV) in fuel feed simulant

Pu(IV) concentration can be quantified over wide range of process chemistry conditions

0.6

0.7

0.6

0.7

542 nm

659 nmR2 = 0.998

process chemistry conditions

0.3

0.4

0.5

bsor

banc

e

0.3

0.4

0.5

sorb

ance

659 nm

797 nm

854 nmR2 = 0.999

R2 = 0.998

R2 = 0.997

0.1

0.2

Ab

0

0.1

0.2Abs

Pu(IV) concentration variable 0.1 to 10 mM

0450 550 650 750 850 950

wavelength, nm

0 2 4 6 8 10 12

Pu(IV) added, mM

Detection limit: Feed composition: 1.3 M UO2(NO3)2 and 0.8 M HNO3

UO2(NO3)2 does not interfere with Pu(IV) measurements

0.08 mM for Pu(IV)using 659 nm band

Vis-NIR spectra of Np(V) and Np(VI) in fuel feed simulant

Np(V) and Np(VI) concentration can be quantified over wide range of process chemistry conditions

0.16

0.18

0.2

y = 0.0588x - 0.0023R2 = 0.99380.2

0.25614 nm982 nm996 nm

Np(V)

over wide range of process chemistry conditions

Np(VI)

0.1

0.12

0.14

orba

nce

y = 0.0508x - 0.0012R2 = 0.9937

0.1

0.15

Abso

rban

ce Np(V)

0.04

0.06

0.08abso y = 0.0088x - 0.0003

R2 = 0.9808

0

0.05

0.00 1.00 2.00 3.00 4.00

Np concentration in Simple Feed, mM

0

0.02

550 650 750 850 950 1050

wavelength nm

Np concentration in Simple Feed, mM

wavelength, nmNp(V): variable 0 to 0.1 mMNp(VI): variable 0 to 0.75 mMFeed composition: 0.3 M HNO3

Np(V): variable 0.05 to 3.3 mMFeed composition: 1.3 M UO2(NO3)2 in 0.8 M HNO3

Chemometric PLS models for quantitative analysis:Pu(IV), Np(V) and UO2

2+ (Raman and vis-NIR)

Use Static spectral database for process monitor model developmentLinear fit with slope = 1 indicates agreement between actual and predicted

Pu chemometric modelPLS model for Pu(IV) using UV-vis data

12

Pu model

Linear fit with slope = 1 indicates agreement between actual and predictedvalues, and successful model performance

1 4UO2 model Np(V) chemometric model

0 5

Np(V) model

y = 0.9997x + 0.0006R2 = 0.9999

8

10

12

cted

odel

y = 0.9986x + 0.0008

R2 = 0.9986

1.0

1.2

1.4

odel

el

y = 0.990x + 0.003R2 = 0.990

0.4

0.45

0.5

edodel

y = 0.990x + 0.003R2 = 0.990

4

6

Pu(IV

), m

M p

redi

cu(

IV)],

mM

, mo

0 4

0.6

0.8

UO

22+],

M m

oU

O22+

], M

, mod

e

0.3

0.35

[Np(

V)],

M p

redi

cte

Np(

V)],

mM

, mo

0

2

0 2 4 6 8 10 12

[Pu

0.0

0.2

0.4

0 0 4 0 8 1 2

[U[U

0.2

0.25

0.2 0.25 0.3 0.35 0.4 0.45 0.5

[N

37

Pu(IV), mM[Pu(IV)], mM0 0.4 0.8 1.2

[UO22+], M[UO22+], M

[Np(V)], M[Np(V)], mM

Proof-of-Concept : Applicability of spectroscopic methods for commercial BWR fuel measurementscommercial BWR fuel measurements

Commercial fuels available at PNNL for demonstrationATM-105, BWR, Cooper Nuclear Power Plant; 18 MWd/kg; low burnupATM-109, BWR, Quad Cities I reactor; 70 MWd/kg; high burnup

Fuel dissolved in HNO3Feed adjusted to 5 nitric acid strengths (0.5 to 5M) and 0.7M U(VI)

Performed batch contact on each aqueous feed with 30 vol% TBP-dodecaneFeed, Organic, Raffinate phases

successfully measured by Raman, Vis-NIR

Excellent Agreement between ORIGENExcellent Agreement between ORIGENcode, spectroscopic, and analytical measurement

ATM-109 Nd Np U PuORIGEN 1.1E-02 4.6E-04 7.2E-01 7.5E-03ICP-MS 8.4E-03 4.7E-04 7.2E-01 9.0E-03Spectroscopic 5.4E-03 3.0E-04 7.3E-01 9.6E-03

38

Spectroscopic 5.4E 03 3.0E 04 7.3E 01 9.6E 03ORIGEN / ICP ratio 1.3 1.0 1.0 0.8ORIGEN / Spectroscopic 1.9 1.5 1.0 0.8

Examples of Vis-NIR spectra: BWR Fuel Feed and TBP/dodecane extraction phase

0.6

M HNO3 5.1M HNO3 3 8

B

1.8

2

M HNO3 5.1M HNO3 3.8M HNO3 2 5

A

high burnup ATM-109 aqueous feed ATM-109 TBP/dodecane phase

0.12ATM 109 Fuel Extract

Pu(IV)

0.3

0.4

0.5

orba

nce

M HNO3 3.8M HNO3 2.5M HNO3 1.3M HNO3 0.3

Pu(IV)Pu(VI)

1

1.2

1.4

1.6

orba

nce

M HNO3 2.5M HNO3 1.3M HNO3 0.3

Pu(IV)Pu(VI)

rban

ce Pu(IV) Pu(IV)Pu(VI) Pu(VI)0.06

0.08

0.10

orb

an

ce

ATM-109 Fuel Extract

Fuel Simulant Extract

rban

ce

Nd(III)

0.1

0.2

abso

0.2

0.4

0.6

0.8abso

Np(V)

abso Np(V)

Nd(III)

0 00

0.02

0.04

ab

soab

sorNd(III)

0550 600 650 700 750 800 850 900 950 1000

wavelength, nm

0550 600 650 700 750 800 850 900 950 1000

wavelength, nmwavelength, nmRed - ATM-109 high burn-up, extractBl Si l t f d t t

0.00

550 600 650 700 750 800 850 900 950 1000

wavelength, nmwavelength, nm

39

Blue - Simlant feed extract

Raman spectra of High-Burnup BWR Fuel, ATM-109

Aqueous phase:ATM 109 feed raffinate and simulant

TBP/dodecane phase:ATM 109 f l d i l t

12

14 Simple Feed fuel simulant

ATM-109 Feed, commercial fuel 2.5

3

ATM-109, commercial fuelUO2

2+

ATM-109 feed, raffinate, and simulant ATM-109 fuel and simulant

8

10

12

resp

onse

ATM-109 Raffinate, commercial fuel

UO22+

NO3-

1.5

2

2.5

resp

onse

Simple Feed, fuel simulant

TBP

NO3- dodecane

2

4

6

Ram

an

offset added for clairity0.5

1

Ram

an

offset addedfor clarity

0

2

750 850 950 1050 1150

wavenumber cm-1

0750 1000 1250 1500

wavenumber, cm-1

wavenumber, cm

Flow Testing with PNNL’s Solvent Extraction Test Apparatus (SETA)

Sixteen 2-cm centrifugal contactors Instrumented with Raman and UV VIS NIR spectroscopic probes

pp ( )

Instrumented with Raman and UV-VIS-NIR spectroscopic probes Computer-controlled operation Vessels mounted on balances accurate tracking of material flow Instrumentation ports, connections points long-range flexibility

41

PNNL’ s centrifugal contactors can be additionally instrumented to support instrumentation evaluation and modeling validation

Cold run of centrifugal contactors: Test on-line Raman and Vis-NIR instrumentation

Cold test conditions

•centrifugal contactors; 2-cm, 3600 rpm•Aqueous phase: 12 ml/min•Organic phase: 18 ml/min

ID HNO3, M NO3, M Nd, mMSoln 1 0 1 0 1 0

Aqueous Phase

•Organic phase: 18 ml/min

Aqueous feed

Soln 1 0.1 0.1 0Soln 2 0.1 0.6 2Soln 3 0.1 1.1 5Soln 4 0.1 2.6 10Soln 5 0.1 4.1 14

30%TBP/dodecaneOrganic Phase

Spectroscopic monitoring ofSpectroscopic monitoring of Raffinate stream only

Raman spectra of Solutions 1-5

3.5E+05

4.0E+05

1.5E+05

2.0E+05

2.5E+05

3.0E+05

man

resp

onse

Soln 1Sonl 2Sonl 3Soln 4

1000 1050 1100

Raman is diagnosticfor nitrate ion

0.0E+00

5.0E+04

1.0E+05

500 1000 1500 2000 2500 3000 3500 4000

Ram Soln 5

ID HNO3 M NO3 M Nd mMAqueous Phase

1000 1050 1100

Aqueous feed

wavenumber, cm-1

Visible Spectra of Solutions 1-5

ID HNO3, M NO3, M Nd, mMSoln 1 0.1 0.1 0Soln 2 0.1 0.6 2Soln 3 0.1 1.1 5Soln 4 0.1 2.6 10Soln 5 0 1 4 1 14

0.25

0.35

nce

Soln 1

Soln 5 0.1 4.1 14

Vis-NIR is diagnosticfor Nd3+ ion

0.05

0.15

abso

rban Soln 2

Soln 3Soln 4Soln 5

-0.05450 550 650 750 850

wavelength, nm

Raman process monitor spectral response of Raffinate streamRaffinate stream

C ld t t l tiCold test solutions

ID HNO M NO M Nd MAqueous PhaseAqueous feed

Soln 5

Nitrate band at 1050 cm-1

ID HNO3, M NO3, M Nd, mMSoln 1 0.1 0.1 0Soln 2 0.1 0.6 2Soln 3 0.1 1.1 5Soln 4 0.1 2.6 10

Soln 3

Soln 4O-H region at 3000-4000 cm-1

Soln 5 0.1 4.1 14

30%TBP/dodecaneOrganic PhaseSoln 1

Soln 2

Soln 3

wash

Wavenumber (cm-1)

PLS model results of Raman and Visible Process Monitoring of Centrifugal Contactor Raffinateg g

Raman Process Monitor, Centrifugal ContactorAqueous phase 0.1M HNO3, variable NaNO3/Nd(NO3)3; organic phase TBP/dodecane

ID HNO3, M NO3, M Nd, mMSoln 1 0.1 0.1 0Soln 2 0.1 0.6 2Soln 3 0.1 1.1 5

Aqueous Phase

3.0

4.0

5.0

M, p

redi

cted

Soln 4

Soln 5mixing region

Aqueous feed

, mea

sure

d

Soln 3 0.1 1.1 5Soln 4 0.1 2.6 10Soln 5 0.1 4.1 14

30%TBP/d dOrganic Phase

0.0

1.0

2.0

0 20 40 60 80 100 120 140

Nitra

te, M

H2O wash Soln 1Soln 2

Soln 3

Nitr

ate,

M

30%TBP/dodecane

Vis Process Monitor, Centrifugal ContactorAqueous phase 0.1M HNO3, variable NaNO3/Nd(NO3)3; organic phase TBP/dodecane

15Soln 5

time, min

Drop in Nd3+ is due to extraction

predicted value5

10

15

Nd,

mM

, pre

dict

ed

Soln 2Soln 3

Soln 4mixing region

raffinate

Nd concentration in:

mM

, mea

sure

d

aqueous feed0

0 20 40 60 80 100 120 140time, min

N

H2O wash Soln 1Nd,

Batch contact distribution experiment for Nd(NO3)3in variable Nitrate with TBP/dodecane

16

8

10

12

14

d R

affin

ate,

mM Feed

Raff inate

2

4

6

Nd

in F

eed

and

00 5 10 15

Nd concentration in Feed, mM

A f d

ID HNO3, M NO3, M Nd, mMSoln 1 0.1 0.1 0Soln 2 0.1 0.6 2Soln 3 0.1 1.1 5

Aqueous feedSlope analysis of 3.2 confirms Nd3+

extraction

Soln 4 0.1 2.6 10Soln 5 0.1 4.1 14

Centrifugal Contactor System Instrumented with Raman and VIS-NIR Spectroscopic Probes

Fiber optics forFiber optics for UV/Vis and NIR

t if l t t

Raman Probes R I t

glove portcover

centrifugal contactors installed within glovebox

47

Raman Probes Raman Input fiber optics

Flow Test: Raman and VIS-NIR Monitoring of Fuel Simulantsg

Raman data VIS-NIR dataVIS NIR data

Np(VI) band at 1220 nm

Np(V) band at 980 nm

Nitrate band at 1050 cm‐1

O‐H region at  3000‐4000 cm‐1

UO22+ band at 

871 cm‐1

HNO3 ‐4

HNO3 ‐5

HNO3 ‐6

UO 6

HNO3 ‐1

HNO3 ‐2

HNO3 ‐3

UO2‐1

UO2‐2UO2‐3

UO2‐4UO2‐5

UO2‐6

Data collected on-line in real-time using running contactor

wash

48

Data collected on-line in real-time using running contactor systemFeed contains variable Np(V, VI), HNO3 and UO2(NO3)2

Chemometric model translation to flow test: Agreement between Measured and Predicted Raman on-line measurementsline measurements

PLS model developed on static measurements successfully predict concentrations of U(VI) HNO and nitrate under flow conditions in real time

Fuel Feed Simulant, M

concentrations of U(VI), HNO3, and nitrate under flow conditions in real-time

0.88

Predicted Concentrations from Process Raman Spectroscopy

total nitrate ID HNO3 UO2(NO3)2 NO3 totalHNO3-1 0.5 0 0.5HNO3-2 1 0 1.0HNO3-3 2 0 2.0HNO3-4 4 0 4.0HNO 5 5 0 5 00 5

0.6

0.7

5

6

7

M

total nitratenitric aciduranyl nitrateExpected values

O3,

nitra

te)

O2(

NO

3)2)

HNO3 -5

HNO3 -6

UO2-5

UO2-6

HNO3-5 5 0 5.0HNO3-6 6 0 6.0UO2-1 6 0.1 6.2UO2-2 6 0.2 6.4UO2-3 6 0.3 6.6UO2-4 6 0.4 6.8

0.3

0.4

0.5

3

4

5

ncen

tration, M

tratio

n, M

(HN

O

ntra

tion,

M (U

O

HNO3 -3

HNO3 -4

UO2-2

UO2-3

UO2-4

2UO2-5 6 0.5 7.0UO2-6 6 0.6 7.2

0

0.1

0.2

0

1

2con

conc

ent

conc

en

washHNO3 -1

HNO3 -2

3

UO2-1

49

0 100 200 300 400

time, min

Conclusions

Methods: Using simulants and BWR Spent FuelC fi d lidit f PNNL’ h t i d li h i b thConfirmed validity of PNNL’s chemometric modeling approach using both static and flow testingDemonstrated Raman spectroscopy for on-line monitoring of U(VI), nitrate, and HNO3 concentrations, for both aqueous and organic phases 3 , q g pDemonstrated Vis/NIR for on-line monitoring of Np, Pu, Nd

Testing Facility: PNNL’s Centrifugal Contactor Apparatus Glovebox installation of 16 spectroscopically instrumented centrifugalGlovebox installation of 16 spectroscopically instrumented centrifugal contactorsPerformance of equipment (contactor and spectroscopic) verified

Technical Challenges gDemonstrate applicability of outlined approach for complex simulants and commercial fuels on-line

interferences from fission products and optically active speciesrange of fuel types and flowsheets

Acknowledgements

US DOE Nuclear Energy Advanced Fuel Cycle Initiative gy y(AFCI), Separations Campaign

US NNSA Office of Nonproliferation and InternationalUS NNSA Office of Nonproliferation and International Security (NA-24)

PNNL’s Sustainable Nuclear Power Initiative (SNPI): Centrifugal contactor instrumentation