1603-212 Freq Domain ARMY

Renewable Energy Testing Center

5301 Price AvenueMcClellan, CA 95652916-929-8001 (p)916-929-8020 (f)www.technikon.us

TECHNIKON

Accelerating the Marketplace for Renewable Technologies

Operated by Funded through the Department of Defensertment of De

US Army Armament Research Development and Engineering Center

Demil and Environmental Technology DivisionEnvironmental Sustainment and Energy Branch

Frequency-Domain Analysis of the PEAT Gasifi er Exhaust Flow

Technikon Report # 1603-212

July 2010

US Army Contract W15QKN-05-D-0030US Army Contract W15QKN-05-D-0030Task 6 RETC, WBS # 2.1.2Task 6 RETC, WBS # 2.1.2

Distribution Statement B: Distribution authorized to US Government agencies only; Critical Technology; 10 Feb 06. Other requests for this document shall be referred to: RDECOM/ARDECAMSRD-AAR-AEE-EATTN: K O’ConnorBldg 355Picatinny Arsenal NJ 07806(973) 724-7540

ii

TECHNIKON REPORT # 1603-212 NAJULY 2010

this page intentionally left blank

iii


Director of Measurement TechnologiesKim Hunter, PE Date

Vice President of OperationsGeorge Crandell Date

Frequency-Domain Analysis of the PEAT Gasifi er Exhaust Flow

Technikon Report # 1603-212

This report has been reviewed for completeness and accuracy and approved for release by the following:

iv



v


Table of Contents

Executive Summary .............................................................................................................1

1.0 Introduction ..............................................................................................................3

1.1 Organizational Background ..............................................................................31.2 Report Organization ...........................................................................................41.3 Test/Study Objectives ........................................................................................4

2.0 Scope of Test/Study ..................................................................................................5

2.1 Report Preparation, Review and Quality Assurance and Quality Control (QA/QC) Procedures/Protocols ..................................................................................5

2.2 Study Outline .....................................................................................................52.2.1 Background ..................................................................................................52.2.2 Mechanical Systems .....................................................................................62.2.3 Flow Measurement System ..........................................................................82.2.4 Data Logging ...............................................................................................92.2.5 Data Processing ............................................................................................9

3.0 Study Results and Discussion .................................................................................11

4.0 Conclusion and Recommendations ........................................................................15

Appendix A The Fourier Transform Method ................................................................ 17

Appendix B Data Acquisition and Signal Conditioning ............................................... 25

Appendix C Acronyms and Abbreviations .................................................................... 35

vi


List of Figures and TablesFigure 2-1 Plasma Gasifi er Schematic of Mechanical System ......................................6Figure 2-4 Differential Pressure Transducer and Power Supply ...................................8Figure 2-5 Pitot Tube Placement in Exhaust Stack .......................................................8Figure 2-6 Measurement System Schematic .................................................................9Table 2-1 Recorded Run Data ......................................................................................9Figure 2-7 Data Processing Flow Chart .......................................................................10Figure 3-1 Particulate Buildup on Small Pitot Tube ....................................................11Figure 3-2 Particulate Buildup on Large Pitot Tube ....................................................11Figure 3-3 Exhaust Flow Rate versus Time 600 Second Plot .....................................12Figure 3-4 Exhaust Flow Rate versus Time 30 Second Plot .......................................12Figure 3-5 Exhaust Flow Rate versus Time 3 Second Plot .........................................12Figure 3-6 Exhaust Flow Rate Power Spectrum ..........................................................13Figure 3-7 Exhaust Flow Rate Power Spectrum – Expanded Scale ............................13Table 3-1 Power Spectrum Peaks ...............................................................................13Figure A1 Sampled Signal ..........................................................................................20Figure A2 Harmonic Component #1 ...........................................................................20Figure A3 Harmonic Component #2 ...........................................................................20Figure A4 Harmonic Component #3 ...........................................................................20 Figure A5 Real Fourier (cosine) Coeffi cients .............................................................21Figure A6 Imaginary Fourier (sine) coeffi cients .........................................................22Figure A7 Power Spectrum of the Original Data Set ..................................................23 Figure B1 Correctly Sampled Signal ..........................................................................27Figure B2 Sampled Signal at Nyquist Frequency .......................................................28Figure B3 Aliased Signal Above Nyquist Frequency .................................................28Figure B4 Aliased Signal above Nyquist Frequency ..................................................28Figure B5 6th-Order Bessel Filter with Multiple Feedback Architecture ....................29Figure B6 555 Integrated Circuit Test Oscillator ........................................................30Figure B7 Low-Pass Filter Test at 2.5 Hz ...................................................................31Figure B8 Low-Pass Filter Test at 9 Hz ......................................................................31Figure B9 Low-Pass Filter Test at 22 Hz ....................................................................32Figure B10 Data Acquisition System Quiescent Power Spectrum ...............................32Figure B11 Operating Power Spectrum with Pitot-Static Ports Blocked .....................33

1


EXECUTIVE SUMMARY

Operators of the plasma energy applied technology (PEAT International) plasma thermal destruction and recovery (PTDR) 100-2 plasma gasifi er have observed instabilities in the unit’s gas fl ow rate since operations began in the summer of 2009. In the winter of 2010 a series of experiments was carried out to quantitatively characterize the modes of fl ow instability inherent in the PEAT design. This analysis was done in the frequency domain using Fast Fourier Transform (FFT) methods. It revealed persistent fl ow fl uctuations at fi ve predominant frequencies. The lowest frequency corresponds to the periodic introduc-tion of feed material into the plasma gasifi er. The second and third frequencies most likely represent “hunting” or surging behavior at the scrubber exhaust fan. Both low-frequency instability and vibration are undesirable from the standpoint of system operation and reli-ability. These issues can be addressed through a redesign of the gasifi er fl ow systems.

2



3


1.0 INTRODUCTION

Technikon, LLC is a privately held contract research organization located in McClellan, California, a suburb of Sacramento. Technikon offers process development and demonstra-tion services to industrial and government clients specializing in advanced technologies in the areas of metal casting, point source emissions and renewable energy.

1.1 ORGANIZATIONAL BACKGROUND

The Renewable Energy Testing Center (RETC) program is based on the concept of testing and validation of renewable energy technologies related to biomass feedstock with a par-ticular focus on biofuels for transportation. Technikon has a demonstration facility located in the greater Sacramento, California region that is being utilized for this initiative. The Renewable Energy Testing Center program focuses on support of relevant and emerging renewable energy technologies in the area of cellulosic waste and biomass to energy and fuel conversion technologies that would support the Department of Defense (DOD) need for compliance to Executive Order 13423 that sets goals for the DOD to increase alterna-tive fuel consumption at least 10% annually.

The development of renewable energy technologies presents a major opportunity for re-ducing US dependence on foreign oil. The DOD, the Department of Energy (DOE) and industry share the goals of reducing energy and fuel costs needed to support transportation, manufacturing and production of electricity. A major roadblock to commercialization of renewable energy technologies is that the smaller manufacturers need a place to demon-strate their pilot units, and validate energy and environmental data. The RETC program fulfi lls this need by supplying the support for developers by being a 3rd party renewable energy testing and validation center with a focus on systems meeting DOD renewable fuel requirements.

The objective of RETC is to provide industry with an independent laboratory for develop-

4


ing and evaluating the performance of renewable energy and renewable fuels technolo-gies with respect to robustness, safety, energy effi ciency, environmental effectiveness and other key performance specifi cations. The RETC, and the oversight of the RETC staff, brings together technology developers, government entities and universities in a facility that provides the tools and services needed to bring renewable energy systems to the com-mercialization phase. It also allows developers to integrate technologies that are needed to supply a complete waste to energy system at an accelerated pace and at a signifi cant cost reduction. Present state and federal grant structures are less fl exible and almost exclude the smaller developers from making applications since they do not have the data needed to get awarded. The RETC fi lls this gap in funding and accelerates renewable energy com-mercialization.

1.2 REPORT ORGANIZATION

The report has been organized in three sections to document the design and execution of the test program. Section 2.0 includes a summary of the test equipment and procedures. Section 2.0 also includes Quality Assurance/Quality Control (QA/QC) procedures and spe-cifi c data sampling and conditioning methods. The results are summarized and discussed in Section 3.0.

1.3 TEST/STUDY OBJECTIVES

The objectives of this RETC Subtask are to: 1) quantitatively characterize the fl ow rate oscillations observed at the PEAT gasifi er exhaust stack; and 2) apply data toward an im-proved understanding of the stability of the PEAT gas fl ow system.

5


2.0 SCOPE OF TEST/STUDY

The scope of this study comprises the acquisition and analysis of exhaust gas fl ow data from the plasma energy applied technology (PEAT) plasma thermal destruction and re-covery (PTDR) 100-2 plasma gasifi er. To accomplish this, time-sampled fl ow data were Fourier transformed into the frequency domain where a harmonic, or spectral, analysis was carried out. The results of the analysis are interpreted in the context of the gasifi er’s mechanical design.

2.1 REPORT PREPARATION, REVIEW AND QUALITY ASSURANCE AND QUALITY CONTROL (QA/QC) PROCEDURES/PROTOCOLS

The resulting data were reviewed by Technikon team members to ensure completeness, consistency with the test plan, and adherence to the prescribed quality analysis/quality control (QA/QC) procedures. Appropriate observations, conclusions and recommenda-tions were added to the report to produce a draft report. The draft report was then reviewed by senior management and comments are incorporated into a draft fi nal report prior to fi nal signature approval and distribution.

2.2 STUDY OUTLINE

2.2.1 BACKGROUND

A PEAT PTDR-100-2 plasma gasifi er was brought to the RETC facility to 1) evaluate its suitability for utilization by the Department of Defense; and 2) to demonstrate its per-formance envelope. It became apparent early in the project that the PEAT system would require substantial electrical, mechanical and safety upgrades before the envisioned test program could take place (that effort is detailed in other reports). In the interim, the sys-tem has been operated in a “non syngas” mode. These operations began in the summer of

6


2009 and fl ow instability in the gas handling system was observed from the outset. During the winter of 2009-2010 an attempt was made to measure the exhaust gas fl ow rate from the PEAT system as part of the air emission permitting process. Pitot-static measurements revealed an erratic core velocity in the center of the exhaust duct where the velocity fre-quently dropped to zero. At this point it was decided a more quantitative understanding of gas fl ow in the PEAT system was needed. An experimental program was begun to analyze the fl ow behavior in the frequency domain using Fourier Transform methods.

2.2.2 MECHANICAL SYSTEMS

The plasma gasifi er is schematically depicted in Figure 2-1. Discrete bags of solid feed material are introduced into the plasma stage through an air lock. The feed material is then theoretically converted into ionized plasma by the energy released from an electric arc. The elements in the plasma recombine in the thermal oxidizer to form synthesis gas which is subsequently cleaned in a wet scrubber. Clean synthesis gas is withdrawn from the system through a centrifugal blower (induced draft fan) and will ultimately fuel a motor-generator set. Alternatively, the synthesis gas can be combusted in a thermal oxidizer using an auxiliary liquefi ed petroleum gas (LPG) burner. The resulting combustion products are cleaned in a venturi scrubber and exhausted to the atmosphere.

Figure 2-1 Plasma Gasifi er Schematic of Mechanical System

ThermalOxidizer

VenturiScrubber

MPV

PT

PT

PC

PC

InducedDraft Fan

GasExhaust

PlasmaGasifier

FeedAir-Lock

Feed

Gasifier Pressure Control Loop

Venturi DifferentialPressure

Control Loop

PT

7


The gas fl ow measurements were made in the vertical segment of the gas exhaust duct us-ing a 1/4 inch diameter pitot tube centered in the duct. The confi guration of the measure-ment system is shown schematically in Figure 2-2 and in the accompanying photographs (Figures 2-3 – 2-5).

Figure 2-2 Flow Duct Schematic

Figure 2-3 Data Logging Cart

Induced draft Fan

Differentialpressure

transducer

To exhaust roof vent

From venturi scrubber

Pitot tube Data loggingsystem

plastic tubing

Pitot signal

Static signal

ram port

static ports

8


Figure 2-4 Differential Pressure Transducer and Power Supply

Figure 2-5 Pitot Tube Placement in Exhaust Stack

2.2.3 FLOW MEASUREMENT SYSTEM

The fl ow measurement system comprises the following:

1) Dwyer A23P Pitot Tube ................ for static and ram pres-sure

2) Dwyer 674-424 Pressure Transducer for differential pres-sure (DP) measurement

3) Ronan X54-600 Power Supply ..... for DP transducer

9


Figure 2-6 Measurement System Schematic

2.2.4 DATA LOGGING

Run data were acquired from the fl ow instrumentation by an IOtech Incorporated DBK60 digital data acquisition system running Daq View software as shown in Figure 2-6. The acquired data is then downloaded to a laptop PC, via Ethernet connection, using an IOtech DakLab 2005 where it is converted to Microsoft Excel® format and processed manually. Typical run data are included in Table 2-1.

Table 2-1 Recorded Run Data

2.2.5 DATA PROCESSING

The fl ow in the exhaust duct was measured as a differential pressure at the center of the duct. The output from the differential pressure transducer was sampled at rates ranging from twenty times per second (20 Hz) to fi fty times per second (50 Hz). When plotted as a function of time, the differential pressure data show large random fl uctuations between zero

ethernet

pitot tube static P.pitot tube dynamic P.

Windows® XPDaq View 9.1.27

Differentialpressure/current

transducer

4-20 mA

stack temp.ambient temp.

DakLab2005

Low-PassFilter

1-5 V

DBK60

0-5 V

MEASUREMENT TRANSDUCER TRANSDUCER

OUTPUT RECORDED

VALUE

Stack Temperature Type K Thermocouple mV °F Ambient Temperature Type K Thermocouple mV °F Pitot tube Differential Pressure (DP) Dwyer 674-444 4-20 mA in H2OFiltered DP Signal Low-Pass Filter See Appendix B 0-5 V Volts

10


and 0.25 inches H2O (the transducer limits). Beyond that, little insight into the sources of instability can be gained by looking at the data plotted as a function of time. For disturbances that are not random but, rather, cycle over time, additional information can be gleaned by mathematically transforming the time data into the frequency domain. This is accomplished via the Fast Fourier Transform (see Appendix A). The result is a data “Power Spectrum”, a term adopted from electrical engineering. The power spectrum shows frequencies where signifi -cant cyclic disturbances are occurring, i.e. process instabilities.

One pitfall that must be avoided when ac-quiring time data for frequency-domain pro-cessing is a phenomenon called “aliasing”. Data become aliased when the sampling rate is too slow to capture the most rapidly fl uctuating disturbances in the process. A power spectrum corrupted by aliasing can-not be recovered or corrected. The only strategies for dealing with aliasing are to: 1) increase the time sampling rate, or 2) re-move the higher frequencies from the input signal prior to data acquisition. In this work a low-pass electronic fi lter was used to at-tenuate signal fl uctuations with frequencies above 10 Hz. See Appendix B for addition-al details.

Figure 2-7 Data Processing Flow Chart

Logged flow rate data

Normalizeraw data

Fast Fourier Transform normalized data

Delete DC (f=0) coefficient from FFT results

Compute Power Spectrum from remaining Real

and Imaginary Coefficients

Plot Results

11


3.0 STUDY RESULTS AND DISCUSSION

For the initial fl ow measurements a 1/8th inch pitot tube was employed (Dwyer model 167-6-/CF A27N). The Pitot tube quickly became plugged by a black particulate residue that was entrained in the exhaust gas. This was unexpected because the pressure sampling point was located immediately downstream from the PEAT wet scrubber, where the gas stream should have been clean.

Figure 3-1 Particulate Buildup on Small Pitot Tube

A larger diameter pitot tube (Dwyer model 674-444) was substituted for subsequent test-ing. This provided a 20 minute sampling window without degradation of the fl ow signal from particulate accumulation.

Figure 3-2 Particulate Buildup on Large Pitot Tube

12


The fl ow-versus-time results for a typical test run are shown in Figures 3-3 through 3-5. The instability of the fl ow rate (measured in inches H2O differential pressure) is evident in each of the plots and particularly so on the 0-3 second graph.

Figure 3-3 Exhaust Flow Rate versus Time 600 Second Plot



0.10

0.15

0.20

0.25

0.30

tack

Flo

w i

nche

s H

2OPEAT Flow Rate vs Time

0.00

0.05

0 120 240 360 480 600

St

Time sec.

PEAT Flow Rate vs Time

0.25

0.30

es H

2O

0.20

Flow

in

che

0.10

0.15

Stac

kF

0.00

0.05

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 300 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

Time sec.


0 20

0.25

es H

2O


0.15

0.20

Flow

in

che

0 05

0.10

Stac

kF

0.00

0.05

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Time sec.

13


The fl ow fl uctuations in the plots appear random. However, transforming these data into the frequency domain reveals periodic recurring disturbances at fi ve distinct frequencies. These are shown graphically in Figures 3-6 and 3-7, and in Table 3-1.

Figure 3-6 Exhaust Flow Rate Power Spectrum

Figure 3-7 Exhaust Flow Rate Power Spectrum – Expanded Scale

Table 3-1 Power Spectrum Peaks

0.0004

0.0006

0.0008

0.001r -

DC

Rem

oved

Flow Transducer Power Spectrum40 Hz Sample Rate

Peak 2

Peak 1

Peak 3

Peak 4

0

0.0002

0 2 4 6 8 10 12 14 16 18 20

RM

S Po

wer

Frequency Hz

Peak 5

0.001

Flow Transducer Power Spectrum40 Hz Sample Rate

Peak 1

0.0008

emov

ed

Peak 2

0.0004

0.0006

er -

DC

Re

0.0002

RM

S Po

we

00.0 0.1 0.2 0.3 0.4 0.5

R

F HFrequency Hz

Peak No. Frequency Hz Period seconds1 0.008545 117 2 0.306396 3.26

3 0.908203 1.10 4 4.06128 0.246

5 8.22144 0.122

14


Peak No. 1 represents a disturbance in fl ow resulting from opening the plasma reactor air-lock to introduce a packet of feed material. This occurs every two minutes during normal operation. It is surmised that Peak No. 2 (and possibly 3) indicates a fl ow “surging” condi-tion in the induced draft fan (see Figure 2-1), possibly aggravated by controller instability in the gasifi er pressure control loops. In fact, both conditions may exist and one process may drive the other. The origin of peaks No. 4 and 5 is not clear. Peak 5 is very broad with low amplitude. It appears to be the second harmonic of peak 4. This suggests a resonant phenomenon is contributing.

15


4.0 CONCLUSION AND RECOMMENDATIONS

The frequency and magnitude of fl ow rate excursions in the PEAT syngas exhaust is an area of concern with respect to the design and operation of the PEAT gasifi er and its air pol-lution control system. The frequency peaks in Table 3-1 appear in all of the power spectra taken during testing.

The propagation of disturbances from the feed airlock throughout the gasifi er is undesir-able, and may contribute to problems with the gasifi cation process, but it is unlikely to be a source of mechanical damage or reduced reliability. Anything that induces continuous surging of the fan, however, will ultimately lead to mechanical failure, reduced reliability and increased maintenance. This will be true for both the rotating equipment and for the system’s other electrical and mechanical components exposed to the effects of vibration.

The fan surging problem could be addressed by replacing the centrifugal induced draft fan with a positive-displacement blower which is not susceptible to surging. However, that would require modifi cations to the gasifi er’s pressure control and syngas storage systems.

With approval from the Army, this report may be given to PEAT International as a recom-mendation for redesign of the fan system to reduce emission and mechanical failure.

16



17


APPENDIX A THE FOURIER TRANSFORM METHOD

18



19


If A.1 described an electrical signal, a0/2 would represent the DC voltage base line and the ans and bns would represent the amplitudes of the contributing wave forms at frequencies “nx”. As the number of terms “n” grow large, the Fourier series becomes equivalent to the function it is approximating.

The value of having a Fourier representation is obvious for analyzing signals, vibrations or any phenomena that repeats periodically. For a century this was done through an arduous series of calculations known as the Discrete Fourier Transform. In the 1960s a special case of the DFT was developed which proved immensely simpler and faster when implemented on a digital computer - with certain restrictions. This Fast Fourier Transform (FFT) algo-rithm was used to process all of the data appearing in this report.

Example – Decomposition of a Data Set into Harmonic Components

The following Figures show a “time sampled” data set composed of a constant (DC) com-ponent with a value of 1.0 and three harmonic components with maximum amplitudes of 1.0, 0.5 and 0.2. The data set is represented by a simple Fourier series with 4 terms:

A.2 1 + sin(t/32) + 0.5 sin(t/32) + 0.2 sin(t/32)

where the data set contains 32 points sampled at a rate of 1 Hertz or 1 sample/second.

Fourier’s Theorem – All Real Functions Represented as Sum of Sines and Cosines

The Fourier Transform method has its roots in Fourier’s Theorem which shows that any mathematical function, or set of observations recorded over time, can be represented by a series of sine and cosine functions. These “Fourier series” have the form:

A.1 a0/2 + a1 cos(x) + b1 sin(x) + a2 cos(2x) + b2 sin(2x) … an cos(nx) + bn sin(nx)

20


Figure A1 Sampled Signal

Sampled Signal

2.5

Sampled Signal

1.5

2.0

ue

1.0

Sign

al V

al

0.0

0.5S

-0.50 5 10 15 20 25 30

Time

0 8

1.0

Harmonic Component #1

0.4

0.6

0.8

-0.2

0.0

0.2

Valu

e-0 8

-0.6

-0.4

-1.0

0.8

0 5 10 15 20 25 30

TimeTime

0 8

1.0


0.4

0.6

0.8

-0.2

0.0

0.2

Valu

e

-0 8

-0.6

-0.4

-1.0

0.8

0 5 10 15 20 25 30

Time

Figure A2 Harmonic Component #1

1


0.5

0Valu

e

-0.5

-10 5 10 15 20 25 30

Time



21


Before the sampled data is processed by the FFT algorithm, it is typically “normalized”. That is, the values of the data points are summed together and then each data point is di-vided by that sum. While it is not required by the FFT, normalizing the time data ensures data sets recorded over different periods of time or at different sampling rates will yield the same values for the Fourier coeffi cients (ans and bns). That is, the FFT will transform all normalized data to the same scale regardless of how it was collected. Simply put, normal-izing converts apples and oranges to apples and apples.

Applying the Fourier Transform and Interpreting the Results – Concept of Negative Frequencies, Spectral Lines

The Fourier transform is a complex transformation. This means that input data consist-ing of real values collected over time are transformed into frequency data consisting of complex numbers which have both real and imaginary parts. The sine coeffi cients bn in the frequency domain are imaginary numbers representing the imaginary part of the transform data at frequency n. The real and imaginary coeffi cients of the data transformed from Fig-ure A-1 are shown in Figure A-5 and A-6.

Figure A5 Real Fourier (cosine) Coeffi cients

Fourier Transform Real Values

1

Fourier Transform Real Values

0.75

s.

0.5

sine

Coe

fs

0.25

Rea

l Cos

00 0.25 0.5 0.75

Frequency Hz

22


Figure A6 Imaginary Fourier (sine) coeffi cients

The single real component appears at zero frequency and represents the constant or “DC” portion of the signal. All the harmonic components that make up the data set are sine functions, therefore, the remaining Fourier coeffi cients are all imaginary. Notice at fre-quencies greater than ½ the sampling frequency – 0.5 Hz in this case – the real coeffi cients are mirror images, and the imaginary coeffi cients are inverted mirror images. In fact, the Fourier transform cannot process component frequencies that exceed ½ the sampling fre-quency and those coeffi cients to the right of 0.5 Hz actually represent negative frequencies beginning at (-)0.5 Hz and working their way back to zero. The physical ramifi cations of negative frequencies are not too signifi cant in this context but the frequency limitation of ½ the sampling rate – called the Nyquist frequency - has important consequences for data acquisition. This is discussed further in Appendix B.

Fourier Transform Imaginary Values

1

Fourier Transform Imaginary Values

0.5

e C

oefs

.

00 0.25 0.5 0.75ag

inar

y Si

n

-0.5

Ima

-1

Frequency Hz

23


Power Spectrum – Power Calculated as Root Mean Square of Harmonic Amplitudes (Spectral Lines).

In the midst of this, the important question is how to best present these complex numbers for ease of interpretation. The most widely used representation is the “power spectrum”. It is obtained by taking the root mean square (RMS) value of the real and imaginary coef-fi cients at each frequency:

A.3 RMSn = SQRT(an2 + bn

2)

which results in a positive real-valued number. The term power spectrum comes from elec-trical engineering where the power contributed by a single frequency is directly related to the RMS voltage of that component. To be rigorous, the RMS values of both the positive and negative frequencies should be added together to reconstruct the amplitude of each contributing frequency. In this report the negative frequencies were discarded since we are only interested in relative amplitudes.

Figure A7 Power Spectrum of the Original Data Set

Power Spectrum

1

0.75

alue

0.5

ed R

MS

Va

0 25

Com

bine

0.25

00 0.25 0.5 0.75

Frequency Hz

24



25


APPENDIX B DATA ACQUISITION AND SIGNAL CONDITIONING

26



27


Data Acquisition and Signal Conditioning

Sampling Rate, the Nyquist Frequency and Aliasing

The highest frequency that can be correctly sampled by a data acquisition system is equal to one half of the sampling rate. This is the Nyquist Frequency. Although frequencies above the Nyquist frequency are still captured by the acquisition system, they are misrepresented in the power spectrum and ultimately they corrupt the entire data set. This phenomenon is known as “aliasing”. An aliased data set is one where high frequency components (above the Nyquist frequency) are folded back into the power spectrum at lower frequencies. It is mathematically impossible to reconstruct the original signal from aliased data.

The actual process of aliasing is best shown graphically. Figures B-1 through B-4 show signals of various frequencies all sampled at 10Hz. The 2Hz signal is acquired with excel-lent fi delity, however, when the Nyquist frequency of 5Hz is reached no signal whatsoever is acquired (Figure B-2). As the signal frequency increases above the Nyquist frequency the apparent frequency recorded is much lower as seen in Figures B-3 and B-4.

Figure B1 Correctly Sampled Signal

2.5

2 Hz Signal Correctly Sampled with a 10 Hz Sampling Rate

2

1 5

2

ude

1

1.5

nal A

mpl

itu

1

Sign

0.5

00 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Time Seconds

28


Figure B2 Sampled Signal at Nyquist Frequency

Figure B-3 Aliased Signal Above Nyquist Frequency

Figure B-4 Aliased Signal above Nyquist Frequency

2.5

21 Hz Signal Captured as 1 Hz Due to Aliasing with a 10 Hz Sampling Rate

2

1 5

2

ude

1

1.5

nal A

mpl

it

0 5

1

Sign

0.5

00 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Time Seconds

2.5

5 Hz Signal Not Captured = Nyquist Frequency with a 10 Hz Sampling Rate

2

1 5

2

ude

1

1.5na

l Am

plitu

1

Sign

0.5

00 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Time Seconds

2.5

11 Hz Signal Captured as 1 Hz Due to Aliasing with a 10 Hz Sampling Rate

2

1 5

2

ude

1

1.5

nal A

mpl

it

0 5

1

Sign

0.5

00 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Time Seconds

29


Low-Pass Filtering

Since there is no way to correct a data set once corrupted by aliasing the only alternatives are to 1) increase the sampling rate or 2) remove all of the higher frequency components from the signal before it is sampled. Some combination of the two is usually required in practice. For the data acquisition portion of this work a 6th-order low-pass active fi lter was constructed with a “cutoff” frequency of 10Hz to attenuate the high frequency components of the signal. The data was acquired at sampling rates ranging from 20-50Hz.

Figure B-5 6th-Order Bessel Filter with Multiple Feedback Architecture

Input Signalfrom

Flow Transducer or

Test Oscillator

Filter Block

+Vcc

-Vcc

+

-

Output Signalto

DBK60

Amplifier Block

+Vcc

-Vcc

+

-stage 2

CA

CB

RA

RB

RC

+Vcc

-Vcc

+

-stage 1

CA

CB

RA

RB

RC

+Vcc

-Vcc

+

-stage 3

CA

CB

RA

RB

RC

Stage CA mF CB mF RA kOhm RB kOhm Rc kOhm

1 0.47 1 18 18 122 0.33 1 21 21 133 0.15 1.5 16 16 19

30


Filter Frequency Testing

The performance of the low-pass fi lter was verifi ed using a simple square-wave oscillator constructed from a standard 555 IC timer. The oscillator could generate pulse trains over a range of 2-20Hz.

Figure B-6 555 Integrated Circuit Test Oscillator

The performance of the low-pass fi lter was tested at several input frequencies. The results for three frequencies are shown in Figures B7-B9. All of the data sets have been normal-ized. The low frequency signal (approximately 2.5Hz) passed through the fi lter without attenuation as expected. In addition to the input frequency at 2.5Hz, the spectrum also

1GND

2TRIG

5CTRL

6THRSH

7DISCH

8Vcc

4RESET

3OUT oscillator

output2-20 Hz

Vcc = 15 Volts20 k

700

47 mF

.01 mF

31


shows power input at about 5Hz and 7.5Hz. These are the 2nd and 3rd harmonics of the primary frequency and may represent a “ringing” phenomenon that occurs when the oscil-lator changes state. As the input frequency approaches the fi lter’s 10Hz cutoff frequency, the signal amplitude begins to experience attenuation as seen in the 9Hz plot. As the input frequency is further increased the attenuation becomes more severe reaching a nearly 100 fold decrease at 22Hz.

Figure B-7 Low-Pass Filter Test at 2.5 Hz

Figure B-8 Low-Pass Filter Test at 9 Hz

0 3

0.4

0.5

0.6

0.7

pow

er -

DC

rem

oved

Filter Attenuation TestTest Input Approx. 2Hz

20 Hz Sample Rate

0

0.1

0.2

0.3

0 1 2 3 4 5 6 7 8 9

RM

S p

Frequency Hz

Filter Attenuation Test Test Input Approx. 9 Hz

20 Hz Sample Rate

0.7

20 Hz Sample Rate

0.5

0.6

Rem

oved

0.3

0.4

wer

-D

C R

0 1

0.2

0.3

RM

S Po

w

0

0.1

0 1 2 3 4 5 6 7 8 9

Frequency Hz

32


Figure B-9 Low-Pass Filter Test at 22 Hz

Background Noise – Quiescent and Operating

After the attenuation performance of the fi lter was verifi ed the complete data acquisition system was assembled on the bench. The “no-signal”, or quiescent, output of the data ac-quisition system was recorded. The quiescent power spectrum in Figure B-10 shows very low recorded power (generally 10-4) down to a very low frequency (0.003Hz). The RMS power recorded below this frequency may be the result of drift in the transducer power supply voltage.

Figure B-10 Data Acquisition System Quiescent Power Spectrum

Filter Attenuation Test22 Hz Test Oscillator Input

50 Hz Sample Rate

0.01

d

p

0.008

C R

emov

ed

0 004

0.006ow

er -

DC

0.002

0.004

RM

S Po

00 2 4 6 8 10 12 14 16 18 20 22 24

Frequency Hz

0.0016

Quiescent Background Power Spectrum

0 0010

0.0012

0.0014

r

0.0006

0.0008

0.0010

MS

Pow

er

0.0002

0.0004

0.0006

RM

0.00000.00 0.02 0.04 0.06 0.08 0.10

Frequency Hz

33


A similar “no-fl ow-signal” test was carried out with the pitot-static system mounted in the exhaust duct and the PEAT gasifi er running. Both the pitot and static ports were blocked to prevent any fl ow data from being transmitted to the pressure transducer. The purpose of the test was to identify and quantify signals making their way into the data set that were unrelated to the exhaust fl ow. This test was run to ensure that mechanical vibrations for example did not bias the fl ow data being recorded. The resulting power spectrum is given in Figure B-11. Like the quiescent spectrum, the power spectrum recorded with the pitot and static tubes blocked. It showed uniform low power “noise” across the spectrum but did not contain any identifi able frequency components that could bias the acquired fl ow signal.

Figure B-11 Operating Power Spectrum with Pitot-Static Ports Blocked

0.0010

Operating Background SpectrumPitot/Static Sensors Blocked

0.0008

0.0010

0 0004

0.0006

MS

Pow

er

0.0002

0.0004RM

0.0000

0 1 2 3 4 5 6 7 8 9

Frequency HzFrequency Hz

34



35


APPENDIX C ACRONYMS AND ABBREVIATIONS

36



37


Acronyms and Abbreviations

DFT Discrete Fourier TransformDOD Department of DefenseDOE Department of EnergyDP Differential Pressure FFT Fast Fourier TransformLPG Liquefi ed Petroleum Gas PEAT Plasma Energy Applied TechnologyPTDR Plasma Thermal Destruction and RecoveryQA/QC Quality Analysis/Quality ControlRETC Renewable Energy Testing CenterRMS Root Mean SquareWBS Work Breakdown Structure

1603-212 Freq Domain ARMY

Documents