Anda di halaman 1dari 28





Seminar Report

Photonic Crystal Fiber




I hereby declare that the seminar report entitled, Photonic Crystal Fiber has been prepared
by me under the guidance and supervision of Dr. R K Madilla during the Academic Session

ID: 2010UEC104
Final yr B.Tech.
Electronics and Communication Engineering.



This is to certify that Siddharth Sethia, Institute ID No. 2010UEC104, VIII Semester, B.Tech.
Electronics and Communication Engineering has successfully completed his seminar work
entitled Photonic Crystal Fiber .
The seminar report is found to be satisfactory and approved for submissions.

Date: Dr. R K Madilla
Asst. Professor
Department of ECE Engg.
MNIT, Jaipur



I take this opportunity to express my deep sense of gratitude and respect towards Dr. R K
Madilla, Assistant Professor, Dept. of Electronics and Communication Engineering,
Malaviya National Institute of Technology, Jaipur. I am very much indebted to him for the
generosity, expertise and guidance I have received from him while working on this seminar
and throughout my studies.
I would also like to thank seminar coordinators Dr. K K Sharma Sir ( Professor, Dept. of
Electronics and Communication Engineering, MNIT, Jaipur)and Dr. M S Ansari Sir
(Assistant Professor, Dept. of Electronics and Communication Engineering, MNIT
Jaipur), who helped me for the completion of my report.

Siddharth Sethia



A novel approach towards load forecasting has been proposed which predicts the data 1h
into the future. The prediction has been done with a moving window manner which is based
on real time data collected from ISO New England. Load Forecasting is an important tool
required in the area generation control and resource dispatch. The data collected is usually
noisy and have complicated load features. The authors of the paper after getting motivation
from their previous work have presented a wavelet neural network with data pre-filtering.
The main idea is to incorporate in the system, is a technique to detect spikes in load data and
correct them. Wavelet decomposition is used to decompose the load profiles into multiple
components at different frequencies; separate neural networks (NN) are applied to capture the
individual features. The features are then combined to form the final forecasts. In order to
employ a moving forecast, 12 wavelet NN are them applied on test results. The wide range of
comparison thus validates the effect of data pre-filtering and the accuracy of wavelet NN
based on the data set collected from ISO New England.



1. Introduction 7
2. Brief History of PCF 8
3. Physical Properties 10
3.1 Effective Refractive Index
3.2 Filling Factor
3.3 Attenuation
3.4 Dispersion
3.5 Non-Linearities
4. Fabrication 14
5. Applications 20
6. Future Research Directions 22
7. Conclusion 24

References 26


Fig.3.1. Scatter plot of the actual load vs. the model 14

Fig.3.2. Correlation between the actual load and the model 15

Fig.4.1. Amplitude Spectrum for 5 min load data 20

Fig.4.2. Three channel filter bank 21

Fig.5.1. Neuron Mathematical Model 24

Fig.6.1. Difference in prediction after wavelet decomposition 27

Fig.6.2. Box plots for forecasting errors for 5 to 60 min outs 29


Photonic-crystal fiber (PCF) is a new class of optical fiber based on the properties
of photonic crystals. Because of its ability to confine light in hollow cores or with
confinement characteristics not possible in conventional optical fiber, PCF is now finding
applications in fiber-optic communications, fiber lasers, nonlinear devices, high-power
transmission, highly sensitive gas sensors, and other areas.
Photonic crystal fibers may be considered a subgroup of a more general class
of microstructured optical fibers, where light is guided by structural modifications, and not
only by refractive index differencesevaluations of various sophisticated nancial products on
energy pricing oered by the market.



The PCF domain started by the idea that light could be trapped inside a 2D photonic crystal
made of silica capillaries running along the fiber length. At that time, the idea was to drill
microscopic holes inside a silica rod, then to draw it down to fiber. However, this technique
could not be used since the instruments were not able to drill into silica. The first successful
PCF was made by stacking several silica capillaries, then drawing the whole structure down
to fiber. One important parameter describing the fiber geometry is the diameter-to-pitch ratio
d/pitch. For this first PCF prototype, the ratio of 0.2 proved to be too small for bandgap
guidance, therefore the central capillary was replaced by solid silica for standard index
Meanwhile, the Maxwell equations were numerically solved and proved that a bandgap could
occur inside a photonic crystal fiber.
(Birks, Roberts, Russel, Atkin, and Shepherd, "Full 2-D photonic bandgaps in silica/air
structures,1995 )
Three years later, the first PCF showing photonic bandap guidance was fabricated using the
stack-and-draw technique. It was only 3cm long, so propagation losses could not be
measured. This figure here shows the field pattern at the fiber exit. Here, the light confined in
the core is composed of two different wavelengths whereas the cladding modes support the
whole input spectrum. So that was the brief introduction to photonic crystal fibers. As I
mentioned, numerical calculations of the guided modes were performed in order to verify that
bandgap guiding was indeed possible in hollow-core fibres.

1970: idea that a cylindrical Bragg waveguide can guide light
1991: idea that light could be trapped inside a 2D PhC made of silica capillaries.
How? By drilling microscopic holes inside a silica rod, then drawing to fiber.
1995: Theoretical proof that bandgap guiding is possible
1996: First PCF prototype


Fig 2.1 Fig2.2

1999: first HC PCF
Photonic bandgap guidance
Stack-and-draw technique
3 cm long!
When illuminated with white light:
Fig 2.3

Fig 2.4 Fig 2.5



There are hollow-core fibres, where guiding is due to bandgap effect. This is the last
generation of hollow-core fibres, made by Crystal Fibers in Danemark. One thing that
prevents photonic crystal fibers from being commonly used in telecommunications is the loss
figure, which is 2 orders of magnitude higher than in standard fibers. Also, that the filling
factor had to be relatively high for a photonic bandgap to appear, and here its value is only

Attenuation: 10 dB/km
SMF = 0.2 dB/km

=1535nm, =100nm
f ~ 10%

= 95%

Fig 3.1


Fig 3.2

There are also solid-core fibers, where guiding occurs by total internal reflection. These
fibers are less subject to variations of the microstructure and are therefore more robust and
more cost-effective than hollow-core fibers. Usually, PCFs are multimode, but a careful
design of the photonic crystal can lead to single-mode propagation for all wavelengths. For
shorter wavelengths, the cladding index increases, therefore the V-parameter stays low and so
does the total number of modes.

Attenuation: 0.5 dB/km

Index guiding -> robustness
Fig 3.3

Fig 3.4 Fig 3.5



A key parameter describing a fiber (and also a waveguide) is the attenuation per unit length.
Here are the attenuation mechanisms. The insertion loss is quite important and depends on
the mode field diameter of the two butt-coupled fibers. There is also the confinement loss,
which is related to the imaginary effective index and decreases very fast with each additional
layer of air holes. Finally, scattering at the glass-air interface accounts for the majority of the
linear loss in most fibers.
Confinement loss:
Propagation loss:
Rayleigh scattering (dopants)

Roughness of the glass-air interface

Impurities (OH
water absorption peaks)
Bending loss:
Lower than in plain silica fibers

Fig 3.6



The nonlinearity of a fiber can be tailored. The effective nonlinearity of a fiber is a function
of the nonlinear refractive index as well as the effective area. Since the minimum achievable
radius is a function of the numerical aperture, extremely small effective areas are achievable
in solid-core fibers, which give huge nonlinearities. In complete contrast, low levels of
nonlinearity in hollow-core fibers make them very attractive for high power delivery

In step-index fibers:
~ 30 (

In solid-core PCF:
~ 550 (

is function of NA: low A

In hollow-core fibers:
~ 0.01 (
-> very attractive for high power delivery




Now, maybe you would like to know how to fabricate these fantastic fibers. Well, the first
step is to obtain a preform, which is a macroscopic version of your photonic crystal fiber. It is
about 2cm wide and can be fabricated by various methods depending on your glass material.
The most used technique when working with silica is capillary stacking whereas drilling and
extrusion are mostly used with soft glasses & polymers.

Fig 4.1



The first step in stacking capillaries is to reduce their dimensions from 1-3cm to ~1mm using
a fiber drawing tower. Then, these smaller tubes are horizontally stacked together in an
hexagonal configuration. A silica rod placed at the center will become the solid fiber core and
several missing capillaries will become the hollow core of the fiber.

Reduce the dimensions of silica capillaries from 1-3 cm to ~1mm.
-> Using a fiber optics drawing tower
Horizontally stack in an hexagonal configuration

Fig 4.2
Fig 4.3

Fig 4.4


As for the extrusion technique, it is used to fabricate the majority of soft glass preforms
(tellurite, chalcogenide, lead silicate, bismuthate,). In this technique, molten material is
forced through a die containing a designed pattern of holes. The die can be made of stainless
steel for example. The extrusion technique was used to fabricate the first nonlinear solid-core
PCF in 2002. It was made of SF6 Schott glass, which has a softening temperature of about
500C. The extrusion technique cannot be used with silica since its softening point is high
enough to melt the die material.
Used to fabricate the majority of soft glass preforms
Tellurite, chalcogenide, lead silicate, bismuthate,
Molten material is forced through a die.
Used to fabricate the first nonlinear solid-core PCF in 2002 (lead silicate glass,
=538C, n=1.805)
Fig 4.5

Fig 4.6

Once you obtain your preform, you draw it down to fiber. Usually two drawing runs are
required to achieve micrometric dimensions. In the first run, the preform is reduced to a cane
of 1mm diameter. Then this cane is introduced inside a jacketing tube and drawn to the final
fiber. One important thing to realize is that the drawing process is not a lithographic process,
where the final structure would be the structure of the preform scaled down. For example, by
changing the pressure inside the holes during drawing, the holes can be enlarged or shrinked.

Preform is heated to 2000C to soften silica
Collapse ratios of ~50,000
run: cane of 1mm diameter
run: cane is introduced inside a jacketing tube + drawn to the final fiber
Not a lithographic process!

Fig 4.7 Fig 4.8


Once the final dimensions are reached, the fabricated fiber is coated with a polymer jacket to
improve its mechanical strength. This is done by applying a liquid polymer (acrylate) to the
optical fiber, then it is solidified by a UV lamp. The final product is then wound onto large
spools and is ready to be shipped to you.


Fig 4.9

Fig 4.10

I will finish this talk by presenting two applications where a lot of research work in being
done. This is the first report of white-light supercontinuum generation in photonic crystal
fibers. A fs pulse was sent into the fiber at 800 nm. The dispersion was engineered to have

the zero dispersion wavelength at the laser wavelength. The small effective area in this all-
silica fiber futher enhanced the nonlinear effects. For more details on nonlinear optics in PCF,
there is this very nice article recently published in Nature Photonics.
First report of white-light supercontinuum generation:
A 100 fs pulse was sent @ =800nm. Peak power=kW
Dispersion was engineered to have ZDW @ =800nm.
Small Aeff to enhance nonlinear effects

Fig 5.1
Since PCF can be filled with gases or liquids, another interesting application is to make gas
cells and to study light-matter interactions. Here is a transmission spectrum of a hollow-core
fiber filled with acetylene gas. This fiber can then be used as a frequency reference.

All-fiber gas cells

Fig 5.2

Fig 5.3


Also, by filling fibers with liquids, temperature sensors can be fabricated. When a certain
wavelength is reached, the effective index of the guided mode reaches that of the liquid and
light becomes evanescent in the core. Since the refractive index of the liquid is a function of
temperature, temperature monitoring can be made.

Temperature sensors
Fig 5.4
Fig 5.5




The idea of this paper is illustrated schematically in Figure 1a, where a microstructured
optical fiber preform is first fabricated on a macroscopic scale and then drawn to the desired
dimension to create the TMC. It is then cleaved and the resulting microstructure is
interrogated transversely. Figure 1b and 1c illustrate the conceptual difference between a
planar device fabricated lithographically and fiber process mentioned above. As can be seen,
similar structures could be placed in a microstructured fiber as could be written into a planar
chip. However, the fiber draw process can produce many such fiber devices at once and the
resulting smoothness of the microstructure would give far superior optical performance over
lithographically written devices [4]. Also, due to the high longitudinal consistency in the fiber
draw process, the fiber devices have the same microstructure along their entire length. This
makes any structure in the fibers semi-infinite in extent, providing the Tall in TMC.

In conclusion, we introduce a novel geometry utilizing fibers transversely as an effective
alternative to planer photonic chip devices. We successfully demonstrate this concept using a
photonic crystal fiber that is dynamically switched using microfluidics.



Fig 5.6




Accurate load forecasting is very important for electric utilities in acompetitive environment
created by the electric industry deregulation. The review of some statistical and articial
intelligence techniques that are used for electric load forecasting has been done. The factors
that aect the accuracy of the forecasts such as weather data, time factors, and customer
classes, as well as economic and end use factors are discussed. Load forecasting methods use
advanced mathematical modeling. Additional progress in load forecasting and its use in
industrial applications can be achieved by providing short-term load forecasts in the form of
probability distributions rather than the forecasted numbers. This paper presents a method of
wavelet neural networks with data pre-filtering to forecast very short-term loads 1 h intothe
future in 5-min steps in a moving window manner. The spike filtering methods remove spikes
in real-time. This WNN method can capture the load components at different frequencies.
Daubechies-4 with two-level decomposition is the best configuration, which balances the
decomposed level, the filter length, and the minimum padding length for decomposition.
Symmetrization is shown to be the best strategy to handle the distortion. Applying the
relative increment transformation to load series enhances the load stationarity. Based on test
results, 12 dedicated wavelet neural networks are used to performmoving forecasts every 5
min. Numerical testing shows accurate predictions with small standard deviations for VSTL
based on the data set from ISO New England.


[1] A.G. Bakirtzis, V. Petridis, S.J. Kiartzis, M.C. Alexiadis, and A.H. Maissis.A Neural Network Short-Term
Load Forecasting Model for the Greek Power System. IEEE Transactions on Power Systems, 11:858863,
[2] B.J. Chen, M.W. Chang, and C.J. Lin. Load Forecasting using Support Vector Machines: A Study on
EUNITE Competition 2001.Technical report, Department of Computer Science and Information
Engineering, National Taiwan University, 2002.
[3] W. Charytoniuk, M.S. Chen, and P. Van Olinda. Nonparametric Regression Based Short-Term Load
Forecasting. IEEE Transactions on Power Systems, 13:725730, 1998.
[4] H. Chen, C.A. Canizares, and A. Singh.ANN-Based Short-Term Load Forecasting in Electricity Markets.
Proceedings of the IEEE Power Engineering Society Transmission and Distribution Conference, 2:411
415, 2001.
[5] M.Y. Cho, J.C. Hwang, and C.S. Chen. Customer Short-Term Load Forecasting by using ARIMA Transfer
Function Model. Proceedings of the International Conference on Energy Management and Power Delivery,
1:317322, 1995.
[6] N. Christiani and J.S. Taylor.An Itroduction to Support Vector Machines and Other Kernel-Based Learning
Methods. Cambridge University Press, Cambridge, 2000.
[7] T.W.S. Chow and C.T. Leung. Nonlinear Autoregressive Integrated Neural Network Model for Short-Term
Load Forecasting. IEE Proceedings on Generation, Transmission and Distribution, 143:500 506, 1996.
[8] J. Dudhia and J.F. Bresch.A Global Version of the PSU-NCAR Mesoscale Model. Monthly Weather
Review, 130:29893007, 2002.
[9] R.F. Engle, C. Mustafa, and J. Rice.Modeling Peak Electricity Demand. Journal of Forecasting, 11:241251,
[10] J.Y. Fan and J.D. McDonald.A Real-Time Implementation of Short-Term Load Forecasting for
Distribution Power Systems.IEEE Transactions on Power Systems, 9:988994, 1994.
[11] E.A. Feinberg, J.T. Hajagos, and D. Genethliou.Load Pocket Modeling.Proceedings of the 2nd IASTED
International Conference Power and Energy Systems, 5054, Crete, 2002.
[12] E.A. Feinberg, J.T. Hajagos, and D. Genethliou.Statistical Load Modeling. Proceedings of the 7th IASTED
International Multi- Conference: Power and Energy Systems, 8891, Palm Springs, CA, 2003.
[13] E.A. Feinberg, J.T. Hajagos, B.G. Irrgang, R.J. Rossin, D. Genethliouand D.E. Feinberg. Load pocket
forecasting software. Submitted to IASTED Journal on Energy and Power Systems.
[14] D.B. Fogel. An Introduction to Simulated Evolutionary Optimization.IEEE Transactions on Neural
Networks, 5:314, 1994.
[15] C.W. Gellings. Demand Forecasting for Electric Utilities.The Fairmont Press, Lilburn, GA, 1996.
[16] T. Haida and S. Muto. Regression Based Peak Load Forecasting using a Transformation Technique. IEEE
Transactions on Power Systems, 9:17881794, 1994.
[17] H.S. Hippert, C.E. Pedreira, and R.C. Souza. Neural Networks for Short-Term Load Forecasting: A Review
and Evaluation. IEEE Transactions on Power Systems, 16:4455, 2001.
[18] K.L. Ho, Y.Y. Hsu, F.F. Chen, T.E. Lee, C.C. Liang, T.S. Lai, and K.K. Chen. Short-Term Load
Forecasting of Taiwan Power System using a Knowledge Based Expert System. IEEE Transactions on
Power Systems, 5:12141221, 1990.
[19] O. Hyde and P.F. Hodnett.An Adaptable Automated Procedure for Short-Term Electricity Load
Forecasting. IEEE Transactions on Power Systems, 12:8493, 1997.
[20] A. Khotanzad, R.A. Rohani, T.L. Lu, A. Abaye, M. Davis, and D.J. Maratukulam.ANNSTLFA Neural-
Network-Based Electric Load Forecasting System. IEEE Transactions on Neural Networks, 8:835846,
[21] A. Khotanzad, R.A. Rohani, and D. Maratukulam.ANNSTLF Artificial Neural Network Short-Term Load
ForecasterGeneration Three. IEEE Transactions on Neural Networks, 13:14131422, 1998.
[22] S.J. Kiartzis and A.G. Bakirtzis. A Fuzzy Expert System for Peak Load Forecasting: Application to the
Greek Power System. Proceedings of the 10th Mediterranean Electrotechnical Conference, 3:10971100,

[23] Y. Li and T. Fang. Wavelet and Support Vector Machines for Short-Term Electrical Load Forecasting.
Proceedings of International Conference on Wavelet Analysis and its Applications, 1:399404, 2003.
[24] V. Miranda and C. Monteiro.Fuzzy Inference in Spatial Load Forecasting.Proceedings of IEEE Power
Engineering Winter Meeting, 2:10631068, 2000.
[25] M. Mohandes. Support Vector Machines for Short-Term Electrical Load Forecasting. International Journal
of Energy Research, 26:335345, 2002.
[26] H. Mori and N. Kosemura. Optimal Regression Tree Based Rule Discovery for Short-Term Load
Forecasting. Proceedings of IEEE Power Engineering Society Transmission and Distribution Conference,
2:421426, 2001.
[27] A.D. Papalexopoulos, S. Hao, and T.M. Peng. An Implementation of a Neural Network Based Load
Forecasting Model for the EMS. IEEE Transactions on Power Systems, 9:19561962, 1994.
[28] M. Peng, N.F. Hubele, and G.G. Karady.Advancement in the Application of Neural Networks for Short-
Term Load Forecasting.IEEE Transactions on Power Systems, 7:250257, 1992.
[29] S. Rahman. Formulation and Analysis of a Rule-Based Short-Term Load Forecasting Algorithm.
Proceedings of the IEEE, 78:805816,1990.
[30] S. Rahman and O. Hazim. Load Forecasting for Multiple Sites: Development of an Expert System-Based
Technique. Electric Power Systems Research, 39:161169, 1996.
[31] S. Ruzic, A. Vuckovic, and N. Nikolic.Weather Sensitive Method for Short-Term Load Forecasting in
Electric Power Utility of Serbia.IEEE Transactions on Power Systems, 18:15811586, 2003.
[32] S.E. Skarman and M. Georgiopoulous. Short-Term Electrical Load Forecasting using a Fuzzy ARTMAP
Neural Network. Proceedings of SPIE, 181191, 1998.
[33] J.W. Taylor and R. Buizza. Using Weather Ensemble Predictions in Electricity Demand Forecasting.
International Journal of Forecasting, 19:5770, 2003.
[34] J.W. Taylor and R. Buizza. Neural Network Load Forecasting With Weather Ensemble Predictions. IEEE
Transactions on Power Systems, 17:626632, 2002.
[35] V.N. Vapnik. The Nature of Statistical Learning Theory.NewYork, Springer Verlag, 1995.
[36] H.L. Willis. Spatial Electric Load Forecasting. Marcel Dekker, New York, 1996.
[37] H.T. Yang, C.M. Huang, and C.L. Huang. Identification of ARMAX Model for Short-Term Load
Forecasting: An Evolutionary Programming Approach. IEEE Transactions on Power Systems, 11:403408,
[38] H.T. Yang and C.M. Huang. A New Short-Term Load Forecasting Approach using Self-Organizing Fuzzy
ARMAX Models. IEEE Transactions on Power Systems, 13:217225, 1998.