Visit https://ebookluna.

com to download the full version and

explore more ebooks

Affine Arithmetic-Based Methods for Uncertain Power

System Analysis 1st edition - eBook PDF

_____ Click the link below to download _____

https://ebookluna.com/download/affine-arithmetic-based-
methods-for-uncertain-power-system-analysis-ebook-pdf/

Explore and download more ebooks at ebookluna.com

Here are some recommended products that might interest you.
You can download now and explore!

(eBook PDF) Power System Analysis and Design 6th Edition

https://ebookluna.com/product/ebook-pdf-power-system-analysis-and-
design-6th-edition/

ebookluna.com

Power System Analysis and Design 6th Edition J. Duncan

Glover - eBook PDF

https://ebookluna.com/download/power-system-analysis-and-design-ebook-
pdf/

ebookluna.com

Intelligent Coordinated Control of Complex Uncertain

Systems for Power Distribution and Network Reliability 1st
Edition Meng Xiangping - eBook PDF
https://ebookluna.com/download/intelligent-coordinated-control-of-
complex-uncertain-systems-for-power-distribution-and-network-
reliability-ebook-pdf/
ebookluna.com

Carbon Capture Technologies for Gas-Turbine-Based Power

Plants 1st Edition - eBook PDF

https://ebookluna.com/download/carbon-capture-technologies-for-gas-
turbine-based-power-plants-ebook-pdf/

ebookluna.com
OFF GRID SOLAR: Photovoltaic solar power system for your
home: An easy guide to install a solar power system in
your home (eBook PDF)
https://ebookluna.com/product/off-grid-solar-photovoltaic-solar-power-
system-for-your-home-an-easy-guide-to-install-a-solar-power-system-in-
your-home-ebook-pdf/
ebookluna.com

Meta-analytics: Consensus approaches and system patterns

for data analysis 1st Edition - eBook PDF

https://ebookluna.com/download/meta-analytics-consensus-approaches-
and-system-patterns-for-data-analysis-ebook-pdf/

ebookluna.com

Paper-Based Analytical Devices for Chemical Analysis and

Diagnostics 1st Edition - eBook PDF

https://ebookluna.com/download/paper-based-analytical-devices-for-
chemical-analysis-and-diagnostics-ebook-pdf/

ebookluna.com

Data-Driven and Model-Based Methods for Fault Detection

and Diagnosis 1st Edition - eBook PDF

https://ebookluna.com/download/data-driven-and-model-based-methods-
for-fault-detection-and-diagnosis-ebook-pdf/

ebookluna.com

Applied Power Quality: Analysis, Modelling, Design and

Implementation of Power Quality Monitoring Systems 1st
edition - eBook PDF
https://ebookluna.com/download/applied-power-quality-analysis-
modelling-design-and-implementation-of-power-quality-monitoring-
systems-ebook-pdf/
ebookluna.com
AFFINE
ARITHMETIC-
BASED METHODS
FOR UNCERTAIN
POWER SYSTEM
ANALYSIS
AFFINE
ARITHMETIC-
BASED METHODS
FOR UNCERTAIN
POWER SYSTEM
ANALYSIS

ALFREDO VACCARO
ANTONIO PEPICIELLO
Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
Copyright © 2022 Elsevier Inc. All rights reserved.

MATLAB® is a trademark of The MathWorks, Inc. and is used with permission.

The MathWorks does not warrant the accuracy of the text or exercises in this book.
This book’s use or discussion of MATLAB® software or related products does not constitute
endorsement or sponsorship by The MathWorks of a particular pedagogical approach or particular
use of the MATLAB® software.
No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher. Details on how to seek
permission, further information about the Publisher’s permissions policies and our arrangements
with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency,
can be found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the
Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and experience
broaden our understanding, changes in research methods, professional practices, or medical
treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described herein. In
using such information or methods they should be mindful of their own safety and the safety of
others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of products
liability, negligence or otherwise, or from any use or operation of any methods, products,
instructions, or ideas contained in the material herein.

ISBN: 978-0-323-90502-2

For information on all Elsevier publications

visit our website at https://www.elsevier.com/books-and-journals

Publisher: Charlotte Cockle

Acquisitions Editor: Graham Nisbet
Editorial Project Manager: Sara Valentino
Production Project Manager: Manju Thirumalaivasan
Designer: Mark Rogers
Typeset by VTeX
Contents

Preface xi
Acknowledgments xiii

1. Uncertainty management in power systems 1

1.1. Sampling methods 2
1.2. Analytical methods 3
1.3. Approximate methods 3
1.4. Non-probabilistic methods 4
References 6

2. Elements of reliable computing 9

2.1. Interval arithmetic 9
2.2. Affine arithmetic 10
2.3. Solving uncertain equations by AA 14
2.4. Reliable solution of non-linear equations 16
2.5. Reliable solutions of constrained optimization problems 18
References 22

3. Uncertain power flow analysis 23

3.1. Problem formulation 24
3.2. Affine arithmetic based solution of the power flow equations 25
3.3. Numerical results 29
3.4. Robust formulation of the power flow equations 39
3.5. Case study 42
References 47

4. Uncertain optimal power flow analysis 49

4.1. Mathematical background 49
4.2. Numerical results 55
4.3. Robust formulation of optimal power flow problems 58
4.4. Case study 62
4.5. Remarks 63
References 63

vii
viii Contents

5. Unified AA-based solution of uncertain PF and OPF problems 65

5.1. Theoretical framework 65
5.2. Applications 70
5.3. Numerical results 72
5.4. Computational requirements 79
5.5. Remarks 79
References 79

6. Uncertain power system reliability analysis 81

6.1. Markov Chains 82
6.2. Uncertain Markov Chains analysis by AA 85
6.3. Case studies 88
References 91

7. Uncertain analysis of multi-energy systems 93

7.1. Optimal scheduling of an energy hub 94
7.2. Case study 98
References 104

8. Enabling methodologies for reducing the computational burden

in AA-based computing 105
8.1. PF analysis 105
8.2. OPF analysis 106
8.3. AA-based computing 106
8.4. Numerical results 108
8.5. Remarks 122
References 122

9. Uncertain voltage stability analysis by affine arithmetic 123

9.1. AA-based calculation of PV curves 128
9.2. Numerical results 129
9.3. Remarks 133
References 133

10. Reliable microgrids scheduling in the presence of data

uncertainties 135
10.1. Deterministic optimization 136
10.2. Robust optimization 138
 Contents ix

10.3. Affine arithmetic-based optimization 138

10.4. Numerical results and discussion 140
10.5. Remarks 143
References 143

Index 145
In memory of my Father (Alfredo Vaccaro)
To my family (Antonio Pepiciello)
Preface

Reliable power system operation requires complex numerical analyses

aimed at studying and improving the security and resiliency of electrical
grids. Many mathematical tools can support power system operators in ad-
dressing this challenging issue, such as power flow and optimal power flow
analyses, static/dynamic security assessment, and power system reliability
analysis. The conventional formalization of these problems is based on the
assumption that the input data are specified by deterministic parameters,
which should be defined by the analyst either from a snapshot of the power
system or inferred on the basis of some assumptions on the analyzed system.
Consequently, the solutions computed by solving these deterministic
problems are rigorously valid only for the considered power system state,
which represents a limited set of system conditions. Thus, when the input
data are uncertain, numerous scenarios need to be evaluated in the task of
computing robust problem solutions.
To address this problem, this Book introduces the basic elements of
Affine Arithmetic-based computing, outlining its important role in uncer-
tain power system analysis.
Affine Arithmetic is an enhanced model for reliable computing in
which the input data and the problem variables are represented by affine
combinations of primitive noise symbols, which are symbolic variables rep-
resenting the independent uncertainty sources affecting the input data (i.e.
exogenous uncertainties), or approximation errors generated during the
computations (i.e. endogenous uncertainties).
After introducing the mathematical foundations of Affine Arithmetic,
its deployment in solving the most fundamental power system operation
problems is presented and discussed. In these contexts the adoption of
Affine Arithmetic-based computing allows formalizing the power system
state equations in a more convenient formalism compared to the con-
ventional and widely used formulation adopted in interval-based Newton
methods. Thanks to this feature a reliable estimation of the solutions hull
can be reliably estimated by considering the parameter uncertainty correla-
tions and the cumulative effect of all the uncertainty sources.
These results can be generalized in the task of solving a generic mathe-
matical programming problem under uncertainty, which can be formalized
by an equivalent deterministic problem, defining a coherent set of min-
imization, equality, and inequality affine arithmetic-based operators. This
xi
 Visit https://ebookluna.com
now to explore a diverse
collection of ebooks available
in formats like PDF, EPUB, and
MOBI, compatible with all
devices. Don’t miss the chance
to enjoy exciting offers and
quickly download high-quality
materials in just a few simple
steps!
xii Preface

Affine Arithmetic-based formulation allows computing feasible and robust

solution enclosures of complex uncertain optimization problems, reducing
the approximation errors, and the estimation over-conservatism.
The deployment of this computing paradigm is applied for solving a
wide range of power system operation problems, as far as optimal power
flow, voltage stability analysis, equipment reliability assessment, and micro-
grid scheduling are concerned.
Finally, to reduce the computational burden of Affine Arithmetic-based
computing, this Book analyzes the role of formal methods for knowledge
discovery from historical power system operation data, which could also
play an important role in optimally identifying the parameters of the affine
forms describing the uncertain input parameters.
To this aim, a data processing technique-based on Principal component
Analysis is introduced for identifying complex features and hidden rela-
tionships potentially describing regularities in the problem solutions, hence
reducing the problem cardinality and the algorithm complexity. Thanks to
this technique it is also possible to define a formal connection between
the principal components and the noise symbols of the uncertain variables,
which furnish an effective method for the optimal identification of the
affine forms.

Alfredo Vaccaro
Acknowledgments

The Authors wish to thank Prof. Claudio A. Canizares who inspired our
research work. Special thanks go to Prof. Kankar Bhattacharya and Dr.
Juan Carlos Munoz for their valuable contributions in applying Affine
Arithmetic-based computing in voltage stability analysis, and Prof. Alberto
Berizzi for his useful insights about the relations between affine arithmetic
and robust optimization.

xiii
CHAPTER 1

Uncertainty management in
power systems
State of the art and enabling methodologies

Power system operators require sophisticated numerical tools aimed at an-

alyzing the current and the expected operation states, guaranteeing secure
and reliable grid operation. In this context, numerical programming repre-
sents the mathematical backbone supporting the most fundamental power
engineering tools, such as Power Flow (PF) and Optimal Power Flow
(OPF) analyses, state estimation, voltage and frequency regulation, and
contingency analysis. These studies could also support the development of
advanced operations tools for integrated systems management, such as reli-
ability and resilience assessment of multi-carrier energy systems, or optimal
energy flow management in distributed energy hubs, which are relevant
applications in the light of the evolution of modern power grids towards
more integrated, smarter, and sustainable systems.
Numerical programming problems in power system analysis are conven-
tionally formalized by deterministic models, whose parameters are assumed
to be exactly known on the basis of several assumptions on the current
power system state. However, both the models and their parameters are in-
trinsically affected by complex and correlated uncertainty sources, which
can be roughly classified in aleatory and epistemic. The former is irre-
ducible, i.e. it has to be accepted and cannot be controlled, whereas the
latter is related to the accuracy of measurements, models, and parameters
characterizing the system.
Uncertainty sources are mostly related to the complex dynamics of
active and reactive power supply and demand, which may vary due to dif-
ferent causes [1]:
• Variable generation mix depending on the market clearing.
• The increasing penetration of non-dispatchable generators based on
renewable energy sources.
• The unpredictable fluctuation of loads due to unexpected events.
• The increasing geographical dispersion of energy generation.
Affine Arithmetic-Based Methods for Uncertain Power System Analysis Copyright © 2022 Elsevier Inc.
https://doi.org/10.1016/B978-0-32-390502-2.00008-9 All rights reserved. 1
2 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

The presence of uncertainties hinders the effectiveness of deterministic

methods. Indeed, they can affect the problem solutions to a considerable
extent. Hence, reliable solution paradigms, which incorporate information
about data uncertainty are required. These paradigms allow to characterize
the uncertainty associated to input data, and to assess its propagation over
iterative calculations, hence providing a certain level of confidence about
the problem solutions, and assessing their sensitivity to changes in input
data uncertainties.
The scientific literature proposes a wide range of methodologies to han-
dle uncertainty in power system analysis [2,3], and the most promising are
summarized in the next sections.

1.1 Sampling methods

Sampling methods consist in generating a set of scenarios, given some
random variables with a defined probability distribution function. Monte
Carlo is the best example of such methods: although it can provide very ac-
curate solutions, it involves a large computational burden, which is due to
the large number of problems that have to be solved, one for each scenario
[4]. This condition prevents its effective utilization for large-scale power
system analysis, where the computational burden could be prohibitively
expensive [5,6].
Due to this reason, much research effort has been put in proposing en-
hancements of Monte Carlo methods, which try to improve the sampling
process at the cost of accepting an increasing level of risk. For example,
[7] integrates Latin hypercube sampling and Cholesky decomposition in the
Monte Carlo method. In [8], a hybrid solution algorithm based on deter-
ministic annealing expectation maximization algorithm and Markov Chain
Monte Carlo is proposed. Correlation between wind variables through an
extended Latin hypercube sampling is presented in [9]. Finally, chance con-
strained programming problems can be employed for solving uncertain
OPF analysis, in which Monte Carlo and deterministic optimization are
combined to handle the stochastic features of the input variables [10].
As previously highlighted, these methods require the definition of a
proper trade-off between accuracy and computational burden, hence they
are useful only when a defined degree of risk can be accepted.
 Uncertainty management in power systems 3

1.2 Analytical methods

Analytical methods require a simplification of the real problem and a certain
number of assumptions in order to make the problem tractable [11]. Some
simplifications can be obtained by characterizing the output random vari-
ables through multi-linearization [12], convolution techniques or Fourier
Transform [13].
An example of these methods is the cumulant method [14,15] or its
combination with the Gram-Charlier expansion [15] or the integration of
Von Mises functions, to handle discrete distributions [16]. Another formu-
lation is presented in [6], through chance-constrained programming model.
However, many of these techniques require the linearization of PF equa-
tions, which are non-linear, and this process generates additional inaccuracy.
The shortcomings of analytical techniques are multiple, as discussed in
[17–20]. One example could be the common assumption of statistical in-
dependence of the input data, or the approximation of the real probability
distribution with an ideal random variable. In this sense, the lack of deter-
minism, which is characteristic of power systems, especially as an increasing
amount of dispersed generators is being installed, can be a problem in the
correct formulation of the probabilistic models. Furthermore, when valida-
tion of these techniques is performed, the assumptions made are often not
supported by empirical evidence.

1.3 Approximate methods

Approximate methods, such as point estimate methods, have been proposed
as possible alternatives to both analytical and sampling methods [21]. These
methods aim at approximating the statistical properties of the variables re-
sulting as outputs of PF and OPF analyses. For example, in case of first or
second order method, the mean and the variance are propagated through
the calculations, by using Taylor series expansion [22].
In case of the point estimation methods, a certain number of determin-
istic problems, in particular m, which is the number of uncertain variables,
has to be solved in order to estimate the statistical moments of the problem
solutions [23]. Extensions of this method are, for example, Hong’s point es-
timate method [24], which are much faster but their accuracy is guaranteed
only for low uncertainty ranges, and the linearization is a good approxima-
tion only around the nominal operation point.
Other approaches are based on primal-dual interior point method [25],
which allows computing the sensitivity of the uncertain parameters with
4 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

respect to the problem solution. Robust design theory can also be used to
approximate OPF solutions in the presence of multiple uncertainty sources
[26].
Finally, approximate methods have their shortcomings as well. For ex-
ample two-point estimate methods are not feasible for analyzing large scale
systems, especially in the presence of a large number of random variables.

1.4 Non-probabilistic methods

The spectrum of available methods for carrying out reliable analyses in
power systems in the presence of uncertainty sources has enriched over the
last decades. The theory of evidence [27] and the theory of possibility [28]
are two examples.
These mathematical formulations are associated to the imprecise hu-
man knowledge about the system, which allows only imprecise estimates
of values and relations between variables [29]. An example in power sys-
tems is related to wind energy production: indeed, wind can be measured
only locally, but its spatial distribution over a large area is difficult using
probabilities. Additionally, weather forecasts provide qualitative informa-
tion, which is difficult to be represented in a probabilistic way. Therefore,
non-probabilistic modeling and simulations can be useful to power system
engineers.
The application of fuzzy-set theory to power system analysis has been
proposed in several papers [30–32]. In this case, both input data and con-
straints are modeled through fuzzy numbers, which are particular instances
of fuzzy sets [33].
Self-validated computing is another way of representing uncertainty. Its
advantage is the possibility of keeping track of the uncertainty propagation
through the calculations. This propagation does not require much informa-
tion about the uncertainty sources affecting the problem variables [34]. The
first example of self-validated models is Interval Arithmetic (IA), which
represents each variable through interval of real numbers, without a proba-
bility structure [35]. A set of operations can be defined over these intervals,
such that the final computed interval is guaranteed to contain the uncertain
value of the output variable which it represents.
IA has been proposed as a tool for handling uncertainty in power systems
by many authors [18,19,36,37]. However, the solution obtained in this case
is affected by the dependency problem and the wrapping effect [34,38],
which could reduce the effectiveness of the method in terms of results
 Uncertainty management in power systems 5

interpretation. This is mainly due to the impossibility of IA to represent

correlations between uncertain variables. These problems require special
structures of the Jacobian matrices (e.g. M-matrices, H-matrices, diago-
nally dominant matrices, tri-diagonal matrices) [39] in order for IA to be
effective in solving power system operation problems, which are commonly
not found in real systems.
To overcome these limitations, Affine Arithmetic-based (AA) methods
for uncertain power system analysis have been proposed. In this paradigm,
each variable is defined by a first degree polynomial, represented by a
central value and a number of partial deviations, each one representing
a statistical independent uncertainty source. In this way, the equations de-
scribing the problem under study can be expressed more conveniently and
the correlation between the uncertain input variables can be effectively
taken into account.
The main advantage of this paradigm is that it is suitable for solv-
ing large-scale problems, both computationally and in terms of accuracy
and robustness of the obtained solution. In [40] and in [41], AA methods
have been adopted to solve uncertain PF and OPF problems, by represent-
ing all the state and control variables by affine forms, without assuming a
probability density function for each of them. This represents a promising
alternative to stochastic information management in smart grids [42].
Several paper explored this idea, started in [43], of the application of AA
to power system analysis. In [44], the effect of network uncertainties was
processed through AA for solving the state estimation problem in presence
of both synchrophasors and conventional measurements.
In [45], the complementarity conditions were used to represent bus
voltage control through an AA-based model, which returned operation in-
tervals for PF variables in the presence of uncertainty. In [46], an uncertain
PF problem is solved, by formalizing it through interval PF and solving a
quadratic programming optimization.
Additional applications and benefits deriving from the application of
AA in power systems have been proposed in [40,41,47–49]. In particular,
some of them are planning and operation in the presence of a large penetra-
tion of renewable energy sources, electric vehicles integration or demand
response. The main advantage of AA over IA derives by its capacity in
narrowing the tolerance interval of the problem solutions, which IA-based
computations tend to over-estimate, especially when integrated in iterative
solution schemes.
6 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

In this book, solution methodologies based on AA are introduced to

solve the most fundamental power system operation problems. AA is an
enhanced model of self-validated numerical analysis, which represents the
variables of interest as affine combinations of primitive variables describ-
ing statistical independent uncertainty sources. These combinations allow
describing the effects of complex and correlated sources of exogenous un-
certainty, and approximation errors involved in the computations.
This methodology can be enhanced by combining it with advanced
techniques for information discovery and data compression, such as prin-
cipal component analysis. In this way, the parameters of the affine forms
describing the input data uncertainty can be rigorously identified on the
basis of historical information, and the cardinality of the considered prob-
lem can be sensibly reduced.

References
[1] G. Verbic, C. Cañizares, Probabilistic optimal power flow in electricity markets based
on a two-point estimate method, IEEE Transactions on Power Systems 21 (4) (2006)
1883–1893.
[2] P. Chen, Z. Chen, B. Bak-Jensen, Probabilistic load flow: A review, in: Proceedings
of the 3rd International Conference on Deregulation and Restructuring and Power
Technologies, DRPT 2008, 2008, pp. 1586–1591.
[3] B. Zou, Q. Xiao, Solving probabilistic optimal power flow problem using quasi Monte
Carlo method and ninth-order polynomial normal transformation, IEEE Transactions
on Power Systems 29 (1) (2014) 300–306.
[4] M. Hajian, W. Rosehart, H. Zareipour, Probabilistic power flow by Monte Carlo
simulation with Latin supercube sampling, IEEE Transactions on Power Systems 28 (2)
(2013) 1550–1559.
[5] C. Su, Probabilistic load-flow computation using point estimate method, IEEE Trans-
actions on Power Systems 20 (4) (2005) 1843–1851.
[6] H. Zhang, P. Li, Probabilistic analysis for optimal power flow under uncertainty, IET
Generation, Transmission & Distribution 4 (5) (2010) 553–561.
[7] H. Yu, C. Chung, K. Wong, H. Lee, J. Zhang, Probabilistic load flow evaluation with
hybrid Latin hypercube sampling and Cholesky decomposition, IEEE Transactions on
Power Systems 24 (2) (2009) 661–667.
[8] H. Mori, W. Jiang, A new probabilistic load flow method using MCMC in consider-
ation of nodal load correlation, in: Proceedings of the 15th International Conference
on Intelligent System Applications to Power Systems, 2009, pp. 1–6.
[9] H. Yu, W.D. Rosehart, Probabilistic power flow considering wind speed correlation
of wind farms, in: Proceedings of the 17th Power Systems Computation Conference,
2011, pp. 1–7.
[10] M. Hajian, W.D. Rosehart, H. Zareipour, Probabilistic power flow by Monte Carlo
simulation with Latin supercube sampling, IEEE Transactions on Power Systems 28 (2)
(2012) 1550–1559.
[11] A. Schellenberg, W. Rosehart, J. Aguado, Cumulant-based probabilistic optimal power
flow (P-OPF) with Gaussian and gamma distributions, IEEE Transactions on Power
Systems 20 (2) (2005) 773–781.
 Uncertainty management in power systems 7

[12] A.S. Meliopoulos, G.J. Cokkinides, X.Y. Chao, A new probabilistic power flow anal-
ysis method, IEEE Transactions on Power Systems 5 (1) (1990) 182–190.
[13] R. Allan, A. da Silva, R. Burchett, Evaluation methods and accuracy in probabilistic
load flow solutions, IEEE Transactions on Power Apparatus and Systems PAS-100 (5)
(1981) 2539–2546.
[14] R. Allan, M. Al-Shakarchi, Probabilistic techniques in ac load-flow analysis, Proceed-
ings of the Institution of Electrical Engineers 124 (2) (1977) 154–160.
[15] P. Zhang, S.T. Lee, Probabilistic load flow computation using the method of combined
cumulants and Gram-Charlier expansion, IEEE Transactions on Power Systems 19 (1)
(2004) 676–682.
[16] L. Sanabria, T. Dillon, Stochastic power flow using cumulants and Von Mises func-
tions, International Journal of Electrical Power & Energy Systems 8 (1) (1986) 47–60.
[17] A. Dimitrovski, K. Tomsovic, Boundary load flow solutions, IEEE Transactions on
Power Systems 19 (1) (2004) 348–355.
[18] F. Alvarado, Y. Hu, R. Adapa, Uncertainty in power system modeling and compu-
tation, in: Proceedings of the IEEE International Conference on Systems, Man and
Cybernetics, 1992, pp. 754–760.
[19] A. Vaccaro, D. Villacci, Radial power flow tolerance analysis by interval constraint
propagation, IEEE Transactions on Power Systems 24 (1) (2009) 28–39.
[20] Z. Wang, F. Alvarado, Interval arithmetic in power flow analysis, IEEE Transactions
on Power Systems 7 (3) (1992) 1341–1349.
[21] M. Madrigal, K. Ponnambalam, V. Quintana, Probabilistic optimal power flow, in:
Proceedings of the IEEE Canadian Conference on Electrical and Computer Engi-
neering, vol. 1, 1998, pp. 385–388.
[22] X. Li, Y. Li, S. Zhang, Analysis of probabilistic optimal power flow taking account
of the variation of load power, IEEE Transactions on Power Systems 23 (3) (2008)
992–999.
[23] G. Verbic, C. Canizares, Probabilistic optimal power flow in electricity markets based
on a two-point estimate method, IEEE Transactions on Power Systems 21 (4) (2006)
1883–1893.
[24] J.M. Morales, J. Perez-Ruiz, Point estimate schemes to solve the probabilistic power
flow, IEEE Transactions on Power Systems 22 (4) (2007) 1594–1601.
[25] A. Mohapatra, P. Bijwe, B. Panigrahi, Optimal power flow with multiple data uncer-
tainties, Electric Power Systems Research 95 (2013) 160–167.
[26] C. Boonchuay, K. Tomsovic, F. Li, W. Ongsakul, Robust optimization-based DC
optimal power flow for managing wind generation uncertainty, AIP Conference Pro-
ceedings 1499 (1) (2012) 31–35.
[27] G. Shafer, A Mathematical Theory of Evidence, vol. 1, Princeton University Press,
Princeton, 1976.
[28] L.A. Zadeh, Fuzzy sets, Information and Control 8 (3) (1965) 338–353.
[29] P. Smets, Imperfect information: Imprecision and uncertainty, in: Uncertainty Man-
agement in Information Systems, Springer, 1997, pp. 225–254.
[30] P. Bijwe, G.V. Raju, Fuzzy distribution power flow for weakly meshed systems, IEEE
Transactions on Power Systems 21 (4) (2006) 1645–1652.
[31] V. Miranda, J. Saraiva, Fuzzy modelling of power system optimal load flow,
in: Proceedings of the Power Industry Computer Application Conference, 1991,
pp. 386–392.
[32] P. Bijwe, M. Hanmandlu, V. Pande, Fuzzy power flow solutions with reactive limits
and multiple uncertainties, Electric Power Systems Research 76 (1) (2005) 145–152.
[33] X. Guan, W.E. Liu, A.D. Papalexopoulos, Application of a fuzzy set method in an
optimal power flow, Electric Power Systems Research 34 (1) (1995) 11–18.
8 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

[34] J. Stolfi, L.H. De Figueiredo, Self-validated numerical methods and applications, in:
Proceedings of the Monograph for 21st Brazilian Mathematics Colloquium, Citeseer,
1997.
[35] R. Moore, Methods and Applications of Interval Analysis, vol. 2, SIAM, 1979.
[36] S. Wang, Q. Xu, G. Zhang, L. Yu, Modeling of wind speed uncertainty and interval
power flow analysis for wind farms, Automation of Electric Power Systems 33 (1)
(2009) 82–86.
[37] L. Pereira, V. Da Costa, A. Rosa, Interval arithmetic in current injection power flow
analysis, International Journal of Electrical Power & Energy Systems 43 (1) (2012)
1106–1113.
[38] M. Neher, From interval analysis to Taylor models—An overview, in: Proceedings of
the International Association for Mathematics and Computers in Simulation, 2005.
[39] L.V. Barboza, G.P. Dimuro, R.H. Reiser, Towards interval analysis of the load uncer-
tainty in power electric systems, in: Proceedings of the International Conference on
Probabilistic Methods Applied to Power Systems, 2004, pp. 538–544.
[40] A. Vaccaro, C.A. Cañizares, K. Bhattacharya, A range arithmetic-based optimization
model for power flow analysis under interval uncertainty, IEEE Transactions on Power
Systems 28 (2) (2013) 1179–1186.
[41] M. Pirnia, C.A. Cañizares, K. Bhattacharya, A. Vaccaro, An affine arithmetic method
to solve the stochastic power flow problem based on a mixed complementarity formu-
lation, IEEE Transactions on Power Systems 29 (6) (2014) 2775–2783.
[42] H. Liang, A.K. Tamang, W. Zhuang, X. Shen, Stochastic information management in
smart grid, IEEE Communications Surveys and Tutorials 16 (3) (2014) 1746–1770.
[43] A. Vaccaro, C.A. Cañizares, D. Villacci, An affine arithmetic-based methodology for
reliable power flow analysis in the presence of data uncertainty, IEEE Transactions on
Power Systems 25 (2) (2010) 624–632.
[44] C. Rakpenthai, S. Uatrongjit, S. Premrudeepreechacharn, State estimation of power
system considering network parameter uncertainty based on parametric interval linear
systems, IEEE Transactions on Power Systems 27 (1) (2012) 305–313.
[45] M. Pirnia, C.A. Cañizares, K. Bhattacharya, A. Vaccaro, A novel affine arithmetic
method to solve optimal power flow problems with uncertainties, in: Proceedings of
the IEEE Power and Energy Society General Meeting, 2012, pp. 1–7.
[46] R. Bo, Q. Guo, H. Sun, W. Wu, B. Zhang, A non-iterative affine arithmetic method-
ology for interval power flow analysis of transmission network, Proceedings of the
Chinese Society for Electrical Engineering 33 (19) (2013) 76–83.
[47] S. Wang, L. Han, P. Zhang, Affine arithmetic-based dc power flow for automatic
contingency selection with consideration of load and generation uncertainties, Electric
Power Components and Systems 42 (8) (2014) 852–860.
[48] G. Wei, L. Lizi, D. Tao, M. Xiaoli, S. Wanxing, An affine arithmetic-based algorithm
for radial distribution system power flow with uncertainties, International Journal of
Electrical Power & Energy Systems 58 (2014) 242–245.
[49] T. Ding, H.Z. Cui, W. Gu, Q.L. Wan, An uncertainty power flow algorithm based on
interval and affine arithmetic, Automation of Electric Power Systems 36 (13) (2012)
51–55.
 Visit https://ebookluna.com
now to explore a diverse
collection of ebooks available
in formats like PDF, EPUB, and
MOBI, compatible with all
devices. Don’t miss the chance
to enjoy exciting offers and
quickly download high-quality
materials in just a few simple
steps!
CHAPTER 2

Elements of reliable computing

Theoretical foundations

2.1 Interval arithmetic

In IA, each variable x is assumed to be “unknown but bounded” in an
interval of real numbers X = [xinf , xsup ], also known as the tolerance of x.
The following theorem is of foremost importance to define operations on
intervals [1]:
Theorem 2.1 (Fundamental invariant of range analysis for IA). ∀  : p →
q , globally Lipschitz with bounded slope. There exists an interval extension
 I : p → q such that:
       
∀ x1 , .., xp ∈ X1 , .., Xp ⇒  x1 , .., xp ∈  I X1 , .., Xp

Extending elementary operations, such as sums or products, to intervals

is trivial. For example, given two intervals X1 = [x1,inf , x1,sup ] and X2 =
[x2,inf , x2,sup ], some basic operations are:
• Sum:
 
X1 + X2 = x1,inf + x2,inf , x1,sup + x2,sup (2.1)
• Difference:
 
X1 − X2 = x1,inf − x2,sup , x1,sup + x2,inf (2.2)

• Product:
X1 · X2 = [ min(x1,inf x2,inf , x1,inf x2,sup , x1,sup x2,inf , x1,sup x2,sup ),
(2.3)
max(x1,inf x2,inf , x1,inf x2,sup , x1,sup x2,inf , x1,sup x2,sup )]

• Division:
     
X1 /X2 = x1,inf , x1,sup · 1
x2,sup , x
1
2,inf
0 ∈/ x2,inf , x2,sup (2.4)

More complex functions can be obtained by composing these primitive

operators [1,2]. Thanks to Theorem 2.1, the interval resulting from these
operations is guaranteed to enclose the real range of function values.
Affine Arithmetic-Based Methods for Uncertain Power System Analysis Copyright © 2022 Elsevier Inc.
https://doi.org/10.1016/B978-0-32-390502-2.00009-0 All rights reserved. 9
10 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

IA has been applied to solve systems of linear [3,4] and non-linear

equations [5], and optimization problems [6,7]. The obtained solutions,
however, can yield unexpected results [8,9], due to the lack of IA in
representing the correlation between uncertain variables, which leads to
the wrapping problem, as illustrated in Fig. 2.1 [9]. This problem can be
demonstrated by considering an harmonic oscillator evolving through the
state space ẋ1 = x2 , ẋ2 = −x1 . At t = 0, it is represented as a rectangle,
whose length and height are parallel to cartesian axes. Since the evolution
of the rectangle leads to another one which is not parallel to the axes, the
evolution of the interval at t requires adding a series of spurious states,
represented by the black region (rotated rectangle) A B C  D . Due to this
reason, the evolution of the interval, in a few iterations, diverges, covering
the entire phase space.
Another example is given by this simple operation:

X1 · (X2 + X3 ) ⊂ (X1 · X2 + X1 · X3 ) (2.5)

The resulting interval is much wider than the true range, and long com-
putation chains may lead to aberrant solutions. This phenomenon requires
adopting special computing techniques [10].
Finally, the “dependency problem” is another limiting aspect of interval
computation, which is a consequence of the definition of the difference
operator (2.2):

X − X = [xinf , xsup ] − [xinf , xsup ] = [xinf − xsup , xsup + xinf ] = 0 (2.6)

This problem is a consequence of the inability of IA in discriminating

relations between uncertainty sources, which are always assumed to be in-
dependent. If the interval variables are independent, then the results are
correct, such as for X1 = [1, 2] and X2 = [1, 2]: in this case, their difference
X1 − X2 = [−1, 3], leads to correct results. On the other hand, if the interval
variables are the same, the overestimation error is extremely large.
To address these limitations, more advanced paradigms, based on AA,
will be discussed in next section.

2.2 Affine arithmetic

AA, which has been introduced in [2], is a self-validated computing method
that allows modeling errors, imprecise data, and truncation errors. The
 Elements of reliable computing 11

Figure 2.1 Dynamic behavior of an oscillator analyzed through IA.

approach originates from IA, but it keeps track of correlations between the
uncertain variables. For this reason, the resulting bounds are much tighter
and the divergence of the solution intervals observed in IA can be avoided
or mitigated [11].
In AA, an uncertain variable x is represented by an affine form, which
is a first degree polynomial:

x̂ = x0 + x1 ε1 + x2 ε2 + ... + xp εp (2.7)

where x0 and xk are known real coefficients representing the central value
and the partial deviations of the affine form x̂, respectively.
The variables εk , called the noise symbols, are defined over the interval
[−1, 1]. The noise symbols represent independent uncertainty sources af-
fecting the problem variables, which can be induced by computation errors
or exogenous uncertain sources.
12 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

Given an affine form, it can be converted to an interval by defining the

radius rad(x̂), the upper bound x, the lower bound x, and the range |X |.


p
rad(x̂) := |xk | (2.8)
k=1


p
x := x0 + |xk | (2.9)
k=1


p
x := x0 − |xk | (2.10)
k=1

|X | := [x, x] (2.11)
Operations on affine forms require the replacement of real operators
with their associated AA-based version. An affine operator (x̂, ŷ) can be
defined to map a generic function (x, y) to the affine domain. This allows
computing an affine form for ζ = (x, y), which is coherent with the input
affine forms (x̂, ŷ).
An useful feature of linear functions of affine forms, is that their affine
representation can be obtained by rearranging all the original noise symbols
into an affine form, as in the following equations:

x̂ ± ŷ = (x0 ± y0 ) + (x1 ± y1 )ε1 + (x2 ± y2 )ε2 + ... + (xp ± yp )εp (2.12)

λx̂ = (λx0 ) + (λx1 )ε1 + (λx2 )ε2 + ... + (λxp )εp (2.13)

x̂ ± λ = (x0 ± λ) + x1 ε1 + x2 ε2 + ... + xp εp (2.14)

Unfortunately, non-linear functions do not allow ζ to be expressed only
in terms of the original noise symbols:

ζ̂ = (x̂, ŷ) = (x0 + x1 ε1 + x2 ε2 + ... + xp εp , y0 + y1 ε1 + y2 ε2 + ... + yp εp )

(2.15)
In this case, a proper approximating affine function should be defined:

 a (x̂, ŷ) = ζ0 + ζ1 ε1 + ζ2 ε2 + ... + εp ζp (2.16)

 Elements of reliable computing 13

this non-affine operator approximates the function (x̂, ŷ) reasonably well
over its domain:

ẑ =  a (x̂, ŷ) + ζp+1 εp+1 = ζ0 + ζ1 ε1 + ζ2 ε2 + ... + ζp εp + ζp+1 εp+1 (2.17)

where the last term represents the residual or approximation error:

e∗ (x̂, ŷ) = (x̂, ŷ) −  a (x̂, ŷ) (2.18)

In order to handle this approximation error, an additional noise sym-

bol εp+1 is added in (2.17), in order to properly bound the corresponding
approximation errors:

ζp+1 > max e∗ (x̂, ŷ) (2.19)

(ε1 ,ε2 ,...,εp )

Furthermore, affine approximations of different functions, such as  a

in (2.16), can be different depending on the required accuracy, and the
maximum acceptable computational burden. The following approximation
represents a decent trade-off between these requirements:

 a (x̂, ŷ) = α x̂ + β ŷ + ξ (2.20)

where the unknown coefficients α , β , and ξ can be identified by the fol-

lowing theorem:
Theorem 2.2 (Chebyshev approximation theorem for univariate func-
tions). Given a bounded and twice differentiable function  , defined in some
interval xI = [xinf , xsup ], whose second derivative does not change sign inside x.
Let  a (x̂) = α x̂ + ξ be its Chebyshev affine approximation in xI . Then:
(xsup ) − (xinf ) (u) + r (u) d(u)
α= ξ= − αu =α
xsup − xinf 2 dx
r (u) = α u + (xsup ) − αχsup

and the maximum absolute error is:

(u) − r (u)
ζp+1 = | |
2
In this case, the optimal coefficients α and ξ can be easily calculated.
Indeed, by considering the line interpolating the points (xinf , f (xinf )) and
dx = α can be obtained.
(xsup , f (xsup )), the slope d(u)
14 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

Given all these properties, AA can be defined as an intermediate

paradigm between Taylor forms and zonotopes, as described in [12], which
is characterized by a set of advantages, lacking from other first order ap-
proximation paradigms.

2.3 Solving uncertain equations by AA

A particular class of non-linear operation is the multiplication between two
affine forms x̂ and ŷ:


n 
n 
n
x̂ ∗ ŷ = x0 y0 + (x0 yh + y0 xh )εh + (xh yj )εh εj (2.21)
h=1 h=1 j=1

This non-affine operation is traditionally approximated by bounding the

second order terms:


n 
n 
n 
n
(xh yj )εh εj ≤ |xh | |yj | = R(x̂)R(ŷ) (2.22)
h=1 j=1 h=1 j=1

where the R(.) is the radius of the affine form. Starting from this result,
a new noise symbol describing the approximation error can be defined as
follows:

n
x̂ ∗ ŷ = x0 y0 + (x0 yh + y0 xh )εh + R(x̂)R(ŷ)εk (2.23)
h=1

A set of uncertain non-linear equations, where the non-linearity de-

rives from the multiplication of affine forms, can be solved by using the
mathematical operators proposed in [13].
In particular, let’s consider the following non-linear equation:

x̂2 = ẑF = 4 + 0.8ε1 + 0.4ε2 (2.24)

The solution of this equation requires computing the central value x0 and
the partial deviations x1 and x2 of the affine form x̂, which shares the same
primitive noise symbols of the known term ẑF , such that:

x̂2 = x20 + 2x0 x1 ε1 + 2x0 x2 ε2 + R(x̂)2 ε3 = 4 + 0.8ε1 + 0.4ε2 (2.25)

 Elements of reliable computing 15

This problem can be solved by using the similarity operator between affine
forms defined in [13], as follows:
⎧
⎪
⎨ x0 = 4
2

2x0 x1 = 0.8 (2.26)

⎪
⎩ 2x x = 0.4
0 2

The solution of this system of equations allows computing the parameters

of the unknown affine form, namely:

x̂ = 2 + 0.2ε1 + 0.1ε2 (2.27)

This affine form represents the solution of the uncertain problem. By ap-
plying the multiplication operator between affine forms, it follows that:

U = x̂2 − ẑF = R(x̂)2 ε3 = ±0.09 (2.28)

where the term U is the affine approximation error, which can be consid-
ered as the equivalent of the rounding error derived by using floating-point
numbers to approximate real numbers. The same paradigm can be general-
ized to solve uncertain programming problems as described in [13].
These results can be applied to solve non-linear system of uncertain
equations and constrained uncertain optimization problems, where the un-
certain equations are represented by the multiplication of affine forms. The
solution of this class of problems is extremely useful in robust power system
analysis, since the power flow equations, the system constraints, and the
cost functions can be properly expressed by products. A more effective so-
lution of this problem can be obtained by considering the following robust
definition:

n 
n 
x̂ ∗ ŷ = x0 y0 + x0 yh + y0 xh + (xh yj )εj εh (2.29)
h=1 j=1

This form of expressing a non-linear function, allows spreading the en-

dogenous uncertainty on partial deviations of the primitive noise symbols.
These variables can be called second order affine forms and they guarantee
the robust inclusion of the real solution set. The application of this formu-
lation to solve generic uncertain programming problems is discussed in the
next section.
16 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

2.4 Reliable solution of non-linear equations

The adoption of the solution paradigm described in the previous section
does not allow to properly consider the effects of the endogenous uncer-
tainty introduced by AA-based computing, but only to estimate a posteriori
the corresponding approximation errors, which provide a rough estima-
tion of the solution robustness. Hence, the affine forms computed by these
approaches can underestimate the real solution bounds, not providing rig-
orous enclosures of the solution sets. To address this complex issue, in this
section the robust solution paradigm introduced in Eq. (2.29) is adopted.
In particular, let’s consider the following non-linear uncertain equation:


n
x̂ŷ = ẑF = zF0 + zFh εh (2.30)
h=1

where ẑF is a fixed affine form, while x̂ and ŷ are the unknown terms.
To solve this problem the original uncertain equation is recast by explicitly
considering the effect of the ‘second order’ noise symbols introduced by
the product operator, as follows:


n 
n 
x̂ ∗ ŷ = x0 y0 + x0 yh + y0 xh + xh yj εj εh = ẑF (2.31)
h=1 j=1

Hence, by matching the central values and the partial deviations of the two
affine forms, we obtain the following system of equations:

x0 y0 = zF0
 (2.32)
x0 yh + y0 xh + nj=1 xh yj εj = zFh ∀h ∈ [1, n]

where the term nj=1 xh yj εj allows to explicitly quantify the effect of the
approximation error on each partial deviation.
This is a fundamental result since it allows quantifying a priori the effect
of the approximation error on the solution sets, demonstrating that, for this
class of problems, each partial deviation is not a number but an uncertain
variable, and more precisely it is an affine form sharing the same primitive
noise symbols of the fixed affine form. A further, and more useful, result
deriving by this mathematical formulation is that it is possible to reliably
satisfy the uncertain equations (2.32), by considering the worst case bounds
 Elements of reliable computing 17

of the approximation error, which can be computed as:


n
 
Uh = ± xh yj  ∀h ∈ [1, n] (2.33)
j=1

Starting from this result it is possible to compute a rigorous enclosure of

the real solution set by considering the robust formulation of the problem
(2.32), which is described by the following set of deterministic equations:

x0 y0 = zF0
   (2.34)
x0 yh + y0 xh − nj=1 xh yj  = zFh ∀h ∈ [1, n]

To assess the benefits of the proposed formalization let’s consider again the
problem formalized in (2.24), which can be recast as follows:
⎧
⎪
⎨ x0 = 4
2

2x0 x1 + x21 ε1 + x1 x2 ε2 = 0.8 (2.35)

⎪
⎩ 2x x + x x ε + x2 ε = 0.4
0 2 1 2 1 2 2

According to (2.34), a robust solution of this uncertain problem can be

obtained by solving the following set of deterministic equations:
⎧
⎪
⎨ x0 = 4
2

2x0 x1 − x21 − |x1 x2 | = 0.8 (2.36)

⎪
⎩ 2x x − |x x | − x2 = 0.4
0 2 1 2 2

which results in the following affine form:

x̂ = 2 + 0.2178ε1 + 0.1089ε2 (2.37)

In order to assess the robustness of this solution, let’s compute the corre-
sponding bounds:

x̂2 = 4 + 0.8713ε1 + 0.4354ε2 + 0.1067ε3

(2.38)
= [2.6933 ± 0.1067, 5.3067 ± 0.1067]

Hence, by considering the worst case, the solution bounds are:

x̂2 = [2.6933 + 0.1067, 5.3067 − 0.1067] = [2.8, 5.2] (2.39)

18 Affine Arithmetic-Based Methods for Uncertain Power System Analysis

which, as expected, match the bounds of the fixed affine form ẑF . On the
contrary, if we compute the bounds of the solution (2.27), we obtain:

x̂2 = 4 + 0.8ε1 + 0.4ε2 + 0.09ε3 = [2.8 ± 0.09, 5.2 ± 0.09] (2.40)

and the corresponding worst case bounds are:

x̂2 = [2.8 + 0.09, 5.2 − 0.09] = [2.89, 5.11] (2.41)

which are an underestimation of the real solution bounds. This important

result confirms the effectiveness of this AA-based operator in the task of
computing robust enclosures of the solution sets.

2.5 Reliable solutions of constrained optimization problems

The adoption of the robust formulation introduced in the previous sec-
tion for non-linear equations solution can be generalized for solving a
special class of uncertain optimization problems, where the data uncer-
tainties are described by affine forms, and the objective function and the
system constraints are expressed by products of uncertain variables. This
kind of problems can be formalized as follows:

min fˆc (yr , x̂s )

yr ,x̂s

ĝm (yr , x̂s ) = ĝmF ∀r ∈ nF , s ∈ nX , m ∈ nE, l ∈ nI

ĥl (yr , x̂s ) ≤ hl

where yr are the real numbers describing the deterministic control vari-
ables, which can be fixed independently from the input data uncertainties
(e.g. the voltage magnitude of a generator equipped with an Automatic
Voltage Regulator), and x̂s are the affine forms describing the uncertain
state variables, which become uncertain due to the effects of the input data
uncertainties (e.g. the voltage magnitude at the load buses).
A robust solution of this problem can be found by minimizing the affine
cost function, satisfying both the equality and inequality constraints for all
the possible instances of the exogenous uncertainties and for the worst case
of the endogenous uncertainties. The overall problem can be formalized by
 Visit https://ebookluna.com
now to explore a diverse
collection of ebooks available
in formats like PDF, EPUB, and
MOBI, compatible with all
devices. Don’t miss the chance
to enjoy exciting offers and
quickly download high-quality
materials in just a few simple
steps!
Other documents randomly have
different content
 “1865.
“Miss Davies has asked me to meet Miss Clough of
Ambleside (who drew up a plan for co-operation among
teachers), Miss Bostock, and other educational ladies. I
cannot help feeling that our new friend, to whom I am so
devoted and grateful, has had greatly to do with my position
lately. It is almost indefinable, but it would seem as if he had
set a stamp on me, so to speak. Certainly the Cambridge
Examination did something—introduced me to him, for
example—but it is only since Christmas that so many little
courtesies have been paid me, officially, I mean. Only one
other person so helped me.”

In some early letters we have descriptions of life at

Bonaly Tower, which indicate the kind of letters she
might have written if life had been less hurried—

“Newcastle-on-Tyne, Jan. 7, 1873.

“I liked Mr. Knox quite as much last Wednesday. He gave
me a hearty welcome, and asked most affectionately for you.
From what Dr. Hodgson says, he is not doing so much on the
Merchants’ Company Schools as he was. I lunched at Mr.
Pryde’s house, and then went with him and Mrs. Pryde to the
‘Women’s Medical Educational Meeting.’ For the first time I
heard Miss Jex-Blake speak; she spoke well. Mr. P. seems
sensible and liberal in his ideas. When you and I were in
Edinburgh, it seems Mrs. P. and two of their children had
scarlet fever, and he himself was in lodgings, away from ‘his
own fireside.’ Mrs. P. is quite ‘advanced,’ and, as her husband
said, ‘is the most refractory parent’ he has to do with. ‘She
was always wanting something’ (he said before her), ‘or not
wanting something else.’ She did not like her girls to learn so
 much writing or sewing, for instance. Their second girl is to
be brought up for medicine. So, you see, Mr. and Mrs. P.
must be advanced.
“One day last week, we, i.e. Mrs. H., Dr. H., and I went to
lunch with Mrs. MacLaren. Mr. M. is Member for Edinburgh,
and Mrs. and Miss M., as you will perhaps remember, are
working for the Women’s Suffrage. I met there Dr. Guthrie’s
youngest son, a very fine young man, who made a strong
impression on me. He is evidently as fine in mind as in
person.”

In speaking of her visits, she had always much to

say of the interesting persons whom she met at
Bonaly, and of the talk she so thoroughly
appreciated, well described under the heading, “The
Professor at the Breakfast-Table,” in Mr. Meiklejohn’s
“Life of Dr. Hodgson,” as—

“the sparkling table-talk, apt illustration, and racy anecdote

with which the doctor enlivened all the time we sat at table.
Without monopolizing the talk, he never allowed it to flag;
and by manifesting the kindliest interest in the sayings and
doings of all, he induced even the shyest to take his part in a
manner that must have astonished him when he came to
look back upon it.”

Mrs. Hodgson, too, had so much grace and

kindness that even this shyest of her guests was
made so much at home as to be “led to imagine that
he must have sat in that particular corner hundreds
of times before, though now for the first time
conscious of it.”
Another of Miss Buss’ letters (Sept. 8, 1874) gives
an account of the place itself—

“Bonaly, Sept. 8, 1874.

“Edinburgh, to me, is full of you! So you have been
constantly in my mind since my arrival here, last Friday
night.
“Bonaly is five miles out of Edinburgh, but, on a clear day,
there is a splendid view of town, castle, and Arthur’s Seat.
Only, a ‘clear day’ is not a common article, for, since Friday, I
have seen little external sunshine, though, inside, there is
plenty. But Mrs. Hodgson herself is confined to bed, and
looks so fragile that a breath might blow her away. We trust,
however, that she ‘has turned the corner,’ as the doctor says
she may be taken into another room to-day....
“This house is beautifully situated in twenty-eight acres of
its own grounds, and there are hills upon hills all round,
except on the Edinburgh side. Two tiny mountain ‘burns,’ or
streams, run through the grounds, with that constant blue
haze over them—a touch of beauty which we got rarely in
the Alps. In these northern latitudes, it seems to me that
there never is the clear, cloudless sky which we know as the
Italian, but there is another kind of beauty—that of the
greyish-blue haze which envelopes everything with a soft
and indescribably beautiful mantle.
“In consequence of Mrs. Hodgson’s health, I left my dear
boy at home, but if he had come, he and George (Dr. H.’s
son) would have been happy together.
 “Mr. Knox is expected here on Thursday. He has been
asked to meet me, and I hope he will come. How much you
and I liked him. Miss Blyth is also invited.
“I am writing in the midst of snatches of talk, which makes
it difficult to know what I am writing, but you will not mind
jerky sentences, with no particular thread of connection?...
“There is a capital article on Woman’s Suffrage in this
month’s Macmillan; it is by Prof. Cairnes, in answer to
Goldwin Smith’s attack. You do not care so much for this
question as I do, so will scarcely feel the same interest in it.
“Is Agnes pretty well? What is she doing, I wonder? Will
you give her my dear love when you write? There is a very
charming letter from Miss Hierta to me, which shall be sent
to you when I know where you are. What a very sweet
woman she is!
“My Hythe holiday was very pleasant; we were such a
large family party. Did I tell you what darlings Frank’s
brothers are? Arthur (six years old) is quite a picture of
infant beauty, with his blue eyes and curly golden hair; and
he says such funny things and makes droll mistakes. He
rushed at me once, saying, ‘Arnie, look at my apostles!’ ‘Your
what?’ ‘Apostles.’ I found he meant fossils!
“Another day he was reading: ‘And she sung a—a
hullabaloo!’ He meant lullaby.
“Then the baby-boy, whose only experience of trees and
green grass is the disused churchyard at Wapping, insisted
on calling every green field and clump of trees a ‘nice
churchyard!’ Was it not pathetic?
“If you are writing to Miss Hopkins, please remember me
most kindly to her, and tell her I congratulate her on Miss
Robinson’s success: no doubt she has largely contributed to
it.
 “I see Miss Robinson has gained her point, and there is
really a Soldier’s Institute at Portsmouth. The military
element is strong at Hythe, in the School of Musketry, and
we can see how it is that the scarlet uniform and gold
trimmings are so popular. To poor people the fine clothes and
certain pay must be very attractive. Contrast the dress and
appearance of an agricultural labourer with that of the
labourer who has enlisted! And then think of the easy life of
the latter. Do not fancy my estimate of soldiers is altered. I
am looking at them from the point of view of the very poor,
to whom to have a soldier son or brother must be a grand
promotion.
“Mr. Knox came here on Thursday. I like him still very
much, and he likes you and me. He asked most kindly after
you. He also sent a copy of ‘A Night and Day on board the
Mars’ to be forwarded to you, which I duly sent off. You
know he is a staunch teetotaller, and is working desperately
in the cause. He said he had known seventy-five men, of his
own position, ruined by drink, and Dr. Hodgson told me
afterwards that this was no exaggeration. The vice of
drunkenness seems to prevail here more than in London, at
least one hears more of it.
“Mr. Knox has nothing now to do with the Company’s
schools, but has given himself up to rescuing boys (I asked
him where were the girls?), and has been violently attacked
for kidnapping them. An absurd charge, of course. I fear he
is not cold and hardheaded, like the typical Scotchman. But,
all the same, I like him whenever I see him.
“How true is what you say about the money matters of
women! But we are breaking through many of these things,
and a later generation of women will know what
independence means. I hope they will use it properly, for,
after all, we cannot be independent of each other. We have
to live in a community.”
 “Bonaly, Sept. 8, 1874.
“Your long and interesting letter has just come, dear
Annie, after one from me to you is written, sealed up, and
put in the post-bag.
“I will read the letter in the Spectator. It seems to me that
Tyndall only says what you say, namely, that science, so far
as he knows, cannot prove God and immortality. But I do not
see why he need have said as much as he did, except that
he is essentially aggressive.
“That people are unjust to him, I admit, and that this
ignorance of his subject and injustice drive him to attack.”

“Bonaly, Sept. 14, 1874.

“I return Mr. S.’s letter, with which Dr. Hodgson was much
amused, as was I. He admired J.’s poem, Vivia Perpetua,
very much, and said how good and sweet it was. He also
begged me to ask her whether she knows the ‘Vivia
Perpetua’ of Mrs. Flower Adams, whose sister Sarah is well
known for her hymns. He thought J.’s little poem might well
do prefixed to the drama by Mrs. Adams.
“My dear Mrs. Hodgson is still in bed, where she lies so
patiently that she is a living lesson to me. It is curious, but
she always makes me feel gentle and soft—a lesson I
constantly need, and no one else produces the same effect
on me. Had I seen her before my interview with E. D., the
latter would not have been frightened at my—what shall I
say?—violence!
“You have a mesmerizing effect on me, but your influence
is quite different—more on the intellectual side, I think. Mrs.
Hodgson is the sweetest, brightest, most fairy-like woman I
have ever known; and the points of contact between her and
me are so many. I have such strong affection and respect for
her husband—he is so clever, and inspires one with a kind of
awe for his knowledge (which is in a line I can follow), his
brilliancy, his wonderful power of expression, his tenderness,
his extreme conscientiousness, and his resource. But no one
would venture to take a liberty with him, and I can well
imagine the respectful awe in which his pupils hold him.
Then the eldest boy is so near Frank’s age, and I have had
so much to do with him that he is very dear to me. The two
little girls are perfectly charming.
“Then the house is full of books, pictures, statues, busts,
etc. Every side of my taste is represented, and the books
especially are always delightful to me. I suppose the
collection of educational works is quite unique. Dr. H.’s
religious views are very independent of theology; but, as I
have said, he is intensely reverent, and respects other
people’s opinions. His popularity with his Class in University
is immense, as I heard on Saturday, and I can well
understand it is so.
“Mrs. H. is one of those women who is absolutely
unselfish. Her unselfishness extends beyond husband and
children, and she can always speak that soft word that turns
away wrath. They are well matched. She is dependent and
clinging, in the best sense, and he is intensely strong....
“I should like some copies of J.’s ‘Lady Jane Grey.’ Will you
give her my love and ask her?
“When I get home I must get a copy of ‘Hertha’ from
Mudie’s. I know there is one there.
“I shall get back (D.V.) refreshed in every way—
intellectually, physically and morally, and spiritually too, I
hope.”
 In 1880 came the end of this bright chapter of her
life. The death of Dr. Hodgson brought back the
sufferings of the earlier loss in 1860, when Mr.
Laing’s death left so great a blank. Between 1875
and 1880 Miss Buss had lost her father, and Mr. and
Mrs. Payne, and now came the death of Dr. Hodgson
and Miss Chessar in the same month, to all of whom
she had been linked not only by the ordinary ties of
life, in more than ordinary strength, but also by very
special sympathy in her personal work.
Extracts from her letters tell their own story. She
and Dr. Hodgson, with Miss Chessar, Miss Caroline
Haddon, Miss Franks, and some others, had gone to
a great educational congress held in Brussels, in
which many of them were to take active part. On
August 21 Miss Buss writes to her sister—

“A very pleasant journey yesterday. The water quite

smooth, and hardly any one ill. We are at present fourteen
people and are shaking down. I am now going to the Bureau
to get my ticket for the Teachers’ Conference, and then to
the Exhibition.”

“Aug. 23.
“I am sorry to tell you that Dr. Hodgson is very ill. He has
had to come to our place, as really he could not be left. I am
now writing for an English doctor. If necessary, I must
telegraph to Mrs. Hodgson, or, if possible, must return with
him to London, telegraphing for her to meet him. It is very
 sad. He thinks it is some heart affection, but no one can tell
till the doctor has been.”

“Aug. 24.
“Dr. H. is so ill that it is feared he will die.
“I have telegrammed to Mrs. Hodgson, but she cannot get
here till to-night at the earliest. I have been praying most
earnestly that he may live to see her. His lungs are
congested, and he breathes just as our father used to do.
“I have now been with him thirty hours, but a most kind
and experienced teacher, Mr. Harris, a friend of Miss
Haddon’s, is chief nurse.”

On August 17, before leaving Edinburgh, Dr.

Hodgson had written to his friend Mr. A. Ireland—

“My courage fails me as the time draws near for going to

Belgium. For the first time in my life the thought of illness
away from home hangs upon me. I have had queer
sensations and pains in the heart.... The educational
conference lasts from the 22nd to 29th inst. I have just
received a huge 8vo. volume of 1000 pages, and 3 lbs. 9½
ozs. in weight, containing preliminary reports for the six
sections into which the conference is divided.”

While in London he consulted a medical man, who

assured him that he was suffering only from
indigestion.
 But the fatigue and heat of travelling brought on
attack after attack of angina pectoris, and on the
evening of August 24 the end came.
Of this terrible three days Miss Buss writes—

“I do not think there has ever been so awful a time in my

life; in other griefs my brothers were by my side, and able to
help. In this, everything has fallen on me, and in a foreign
country, too. Had it not been for Miss C. Haddon and Mr.
Harris it would not have been physically possible for me to
bear what I have had to go through. Also the girls of my
party were very helpful.
“Dear Mrs. Hodgson does not, as she says, at present
understand things. It is a dream to her: she arrived just
twelve hours too late.
“It is too real to me to be a dream; his dear voice is still
sounding in my ears; he was so patient and so grateful,
thanking us all each time we gave him seltzer-water, etc.
“But I had no idea of death till within a few hours of the
end.
“I was with him just thirty-seven hours. He called for me
at five o’clock on Monday morning. I went at once and gave
him some brandy, and then sent for the doctor while Miss
Chessar stayed with him.
“Dearest mother, I long to have you and my boy safe in
my arms—to make sure of you both.
“How I loved my dear friend no words can express. How
glad I was to have him as my guest, and to travel with him!
Such an opportunity had never occurred before.”
 From this date some part of Miss Buss’ holiday was
always spent with Mrs. Hodgson, whose own words,
after her friend had been taken from her, show what
this friendship was to her also,

“You ask me to tell you something of my friendship with

Miss Buss. I could only do so by giving you a long list of
kindnesses received from her, kindnesses which made one
wonder how a woman leading such a busy life could
remember such things as birthdays, not only of one’s own,
but of one’s children and grandchildren, none of whom were
ever forgotten. The terrible anxiety she went through at
Brussels in 1880, during the Educational Congress there,
must have told heavily on her nerves, already sorely taxed.
My husband went with her to Brussels, and when she found
him ill and suffering at his hotel, she took him to her
lodgings and gave up one of her rooms, which at that time
were very difficult to get, Brussels being very full, and
devoted herself to nursing him night and day for the short
and fatal illness. I can never tell you of all she went through
to help me, but can only say that when we arrived at Bonaly
Tower, near Edinburgh, where we brought our beloved, she
was very ill, the result of what she had gone through, not
only to nurse her old friend, but when all was over to help
and comfort me, utterly forgetful of self. Ever since that sad
time she has been more than a sister to me. I fear now I
took advantage of her wonderful goodness, her wise
judgment, her strict sense of justice, her unselfishness, and
learned more and more to consult her, who was the friend
and helper of all who stood in need of help. To me her loss is
irreparable, and I believe I am only one of a great many who
went to her in times of trouble.”
 But this sorrowful experience was not the only
grief of that year, for Miss Chessar never left Brussels
again, surviving Dr. Hodgson less than a month. She
had not been strong, but no one had in the least
anticipated anything serious, and this second blow,
following so closely on the first, greatly affected Miss
Buss, who thus lost by one stroke the two persons
who were the greatest help and strength in her
work. Like herself, they were both teachers of
remarkable power, and the three friends had set
themselves to raise the general standard of teaching,
while at the same time their sympathies in other
directions cemented a close friendship.
The force of this double loss is given very clearly in
the replies from Mrs. Grey and Miss Shirreff to letters
from Miss Buss, these letters themselves not being
attainable—

“Meran,
“Sept. 20, 1880.
“My dear Miss Buss,
“It was only yesterday that we heard, from Miss
Brough, of the death of Miss Chessar, and I write in both our
names to express our deep and affectionate sympathy with
you in this second, and, I fear, even heavier loss, coming so
soon after Dr. Hodgson’s death. Our own sense of loss is very
heavy; though we knew her so little in private life, she had
inspired us with real and warm personal regard, besides
admiration for her remarkable powers. We are anxious that a
fitting obituary notice should appear in the Journal, if it is not
already done, and have written to Miss Brough to get it
done. Will you help her to do full justice to your common
friend? And please, whenever you have a moment’s leisure,
let us hear how you are yourself.
“It grieves us to hear how your sorely needed holiday has
been turned into a day of sadness and mourning by these
two deaths. Dr. Hodgson’s must have been such a terrible
shock, and from its circumstances have brought upon you so
much to try you, in addition to the personal loss. We women
have lost in him a friend such as we shall not see again, and
he was one of the few left in this dull generation who could
fight with wit as well as earnestness, and had always a good
story to clinch an argument.
“I cannot hear of all the good work going on without a
pang at being so unable to join in any of it, and all my
idleness and care of my useless self has not brought me any
nearer, that I can see or feel, towards ever joining in it
again!... We go to Florence and then to Rome, where I hope
we shall see you in the Christmas holidays. With love from us
both, ever, dear Miss Buss,
“Your affectionate
“Maria G. Grey.”

In November, Mrs. Grey writes again—

“Your letter made us very sad. The loss of two such friends
as Dr. Hodgson and Miss Chessar coming upon you under
such circumstances, and so close together, was enough to
break you down utterly, but, as you do not mention your
health, we trust it did not suffer. We cannot help hoping that
the distressing effect will have worn away enough to let your
 old elasticity of spirits and love of Rome restore you, and
that we may yet have the pleasure of welcoming you here at
Christmas.”

Miss Buss had written to say that Rome was not

possible for this year, and in response Miss Shirreff
speaks of one part of her letter—

“How true is what you say of the terrible void in one’s life
from the loss of early friends, but, believe me, dear Miss
Buss, later friendships may become very close and dear, and
you are far indeed from having overpast the age for making
them. Those to whom mental sympathy has always been the
strong, if not the strongest, link in friendship, have in this
case a great advantage over others, because, while we
outlive other and lighter needs of our nature, the need for
mental companionship never is lost, and this enjoyment can
never cease to give, after close affection, the truest zest to
life. It is therefore never too late to meet with it, though we
become slower in discerning it when it exists. But you have
not reached that point, and with the full vigour of mental
faculty you are ready to seize the full enjoyment of what
responds to your own nature. In hours of sorrow we are so
apt to feel the burden of years that we acquiesce too readily
in the privations they seem to bring.
“I hope your quiet holiday-time spent with your old friend
will send you back strengthened and hopeful to your work. I
cannot express how much we feel your goodness in having
added to it the guidance of this new school (the Maria Grey
Training School) through its difficult early years. Mrs. Grey
joins in love, and says she will write another day.
“Ever affectionately yours,
 “Emily A. Shirreff.”

The visit to Mrs. Hodgson during the holidays did

much to comfort them both, and to strengthen the
bond that never relaxed to the end. The very latest
pleasure of Miss Buss’ life, in the bright interval that
preceded the fatal illness, was a visit at Myra from
this loved and loving friend.
 CHAPTER V.

REST.

“One who never turned his back, but marched

straightforward; Never doubted clouds would break; Never
deemed, though right were worsted, wrong would triumph;
Held we fall to rise, are baffled to fight better, Sleep to
wake!”
R. Browning.

Strangers might easily receive the impression that

Miss Buss was one of those happy persons who,
being blessed with an iron constitution, do not know
what illness means. This was, however, very far from
the fact; for with a temperament so intensely
sensitive, she was in reality one of the women who
can be as ill as they choose to be; and a good deal
of her apparent vigour lay in the strength of the will
which elected not to be ill. “Great minds have wills,
where feeble ones have wishes.” It was just because
she so well knew what could be done by self-control
that she exacted so much self-control from all around
her. From experience she knew how largely the body
may be made the instrument of the spirit, and for
much of her time she kept going by sheer force of
that indomitable will.
It was because she carried this effort too far, in
exacting from her woman’s strength the work that
might have contented several strong men, that she
grew old before her time, and finally broke down,
paying the price of overstrain for some years before
the end came.
All that we can hear of her early life gives the
impression of perfect temper, of unfailing composure,
of unbroken self-command. It is only in later years,
when her great work was completed, that we find
the nervous irritability that is the price paid for over-
work, or, more truly, of over-worry, since it is not
work that kills, but worry.
So much did all around her rely on her strength
and vigour that it is with surprise we note the
recurrence in her letters of such passages as these,
even so many years ago:—

“September, 1872.
“It is simply sickening to think of the crowds who come to
me, and I have been so ailing in health that I have only
managed to get along at all by sitting with Berlin woolwork in
the evening, going to no meetings, and getting to bed at ten
 o’clock. Also, though to tell you this is dreadful, I have got
through this week only on champagne twice a day, with
doses of iron!
“The champagne has, I trust, done its work and set me
up, so I hope to go on without any more until next time! My
throat has been affected without intermission this term, and
the sleepless nights have almost driven me to opiates or to a
doctor. But I think I am better, and the holidays are coming
near.
“This is the history of every term, however, and the
question will arise, how long such a strain can be borne? I do
my best to keep in health, but over-strained nature will have
her way sometimes. This is perhaps a new light on my inner
life. But, my dear Annie, remember every one thinks I am a
proper person on whom to make claims....”

This inability to meet claims to which she would so

gladly have given full space was a very wearing part
of the overcrowding of her life. Here is a regret that
she was compelled to seem to neglect a friend for
whom she would have done anything in her power:—

“Her letter pains me, in a sense, because I know how

heavy is the trial of waiting and doing nothing when there is
the will to work. If only I had some leisure I might go to her
and talk with her.
“But I can give nothing except to those who can come to
me, and not always, or even often, then. Do not say
anything. As the work goes on, we may see a way to keep
her interested in, and cognizant of, our part of it.
 “I had no idea of how much she had cared for me in the
past days, and it is very touching to know it.”

“March, 1873.
“... I hope you have not been thinking harshly of me for
not answering your note or calling, but if you have, you must
in imagination take my place, which is at all times fit to be
occupied by ten ordinary women, but which, at the end of
the school year, with all the examinations and prizes, is large
enough for twenty.”

“December 9, 1873.
“I am going to bed now (eight o’clock), and hope to be
better for a night’s rest.
“Here I am again a prisoner in my room! A sore throat is
the main cause....
“But I am generally out of sorts. I am learning that I
cannot do as I used, and that body will dominate mind and
will.
“I fear you are no better. You had my news? It seems to
me quite foolish for me to be ill and unable to do my work
when the path became suddenly clear, and all so quiet too!...
“Dearest Annie, my love to you. Lately I have often
seemed to want you, but I have never been so long and so
completely broken down—except there was organic disease,
when I had fever—as I have this term, and therefore unable
to go to you.
“There is a lecture at the College of Preceptors to-morrow
night, on ‘English as a Means of Philological Instruction,’ by
Dr. Morris—the Morris. 7.30. Could you go? If so, could you
 join me here a few minutes before seven? Only Miss Fawcett
is going.
“I am better in myself, but cannot yet stand upright or
walk about. Patience is teaching me a great lesson, and I
hope I am learning it, in part, at least.
“... I really think there have never been so many petty
worries crowded together.
“It is all very well for men to say ‘never mind.’ However,
what is to be will be, and strength comes with the need.
“I am much better in health. Why, do you think? I went on
Saturday to my uncle’s perfectly quiet house, and out of the
48 hours slept 25!—2½ hours each afternoon, and 10 hours
each night.
“I am feeling so much better to-day—I slept well last
night. But one of the distressing signs of over-work is
disturbed and light sleep, and my brain is so constantly at
work in day-time that I need deep sleep. So cause and effect
act and react.
“My heart has been wrung too by Mr. Payne’s death. Life
seems so full of anguish as one gets older, that at times I
seem to have no power of being bright and cheerful.”

In addition to the regular work of the school, and

all the claims of outside work and of pupils and
friends, there was a large amount of wear and tear
inevitable in any undertaking on so vast a scale.
There was also much that was painful connected
with the success of the public movement, so far as it
affected small private schools or the work of ordinary
governesses, who all seemed to urge some moral
claim to compensation. It was impossible for the kind
heart not to suffer even when the clear head denied
the validity of the cause of the suffering, as in this
letter in reference to one such case:—

“I wonder dear A. does not remember that when a man

makes a new invention, and thereby ruins many individuals,
he is not expected to compensate them.
“They suffer in the interests of the greater number, and, if
wise, direct their efforts towards working the new invention
or improving on it. This may seem cruel, but it is not so in
the end. There is no reason, human or divine, why A. B. C.,
etc., should put aside a direct benefit to themselves and
others in order to prevent Z. from turning his attention to
some other field of work than that he already occupies. It is
certain that three hundred girls in one school want as much
teaching as thirty girls in ten schools—only they want
different teaching.
“Moral—the big school displaces labour, but does not crush
it.”

In the mere fact of success itself there was trial

enough in many ways. The intensity of her feeling
might be sometimes out of due proportion to the
cause of suffering, but none the less did she suffer
acutely. At the time of greatest triumph—the opening
of the new schools in 1879—there chanced to be one
example which gave rise to an outbreak of
indignation on her part, letting us see how much had
hitherto been hidden even from her friends. Of this
incident she writes—

“It is of no use to try to please people! I do not mean to

try. I will do what seems to me right, and then learn to be
content to be abused, if I can! What with every one’s
‘claims,’ and with people’s ‘rights’ to a seat, always the best!
—friends, family, parents, old pupils, etc., it is all the same!
Every one is dissatisfied, do what one will; some one else is
preferred, some one is neglected.... And so the stings go on,
till I nearly break down under the wounds they inflict. When
barely able to get about again through the work, I hear of
my neglect, etc., of one to whom, in my heart of hearts, it
never occurred to me as possible that any one could accuse
me of ingratitude.
“Pray forgive me, dear Annie, but you can never know the
bitter price one pays for success. I think it as heavy as that
of failure! This has stirred up a depth of scorn and anger of
which I feel ashamed, though I feel almost ashamed, too, of
the race of beings to which I belong.
“I do not know whether it will do any good to have it out,
so to speak, with you. I fear perhaps it will worry you. But as
I have written it, it shall go, and I hope you and I shall meet
next Saturday, when the keenness of the stroke has passed.
I do not, however, think that just now I can write to our
friends. I should not wish to pain them, so silence will be my
best refuge. Do not please say anything. I will fight my fight
out with myself alone.
“God’s law of compensation comes in; He will neither
suffer one to be unduly elated nor depressed.
“It is part of our discipline in life that we should constantly
fail, and I earnestly hope that I may be permitted to try and
try again.
 “But the old days have gone, and it would be better as
well as easier for me for no visitors to be allowed to enter
except the few on the platform and the mothers of girls
taking prizes high in the school.
“Trying to please every one, and to recognize his or her
rights, is not of the least use. Like the miller in the fable, one
only succeeds in pleasing no one.
“There is so much to be grateful and thankful for that I am
really ashamed of myself for feeling vexed. I have not told
you half the vexations to which people subject me, certainly
not because I ignore them, but because by trying to please it
seems impossible to succeed.”

Earlier in this “year of triumph” there is a pathetic

little note to her sister, showing how much stronger
was the “domestic” than the public woman in her—

“February 18, 1879.

“Dearest little Mother,
“Don’t be unhappy, but you did not think how much I
miss your loving little hug and petting.
“No one pets me but you, and occasionally Mrs. Bryant.
Darling boy allows me graciously to pet him, but he does not
make advances to me.
“I want you sometimes, if only to look at!
“Where are we to go at Easter? I was thinking of Hastings.
Let me know.
“Your very loving old
“Arnie.”
 It is not necessary to say that no change really
took place in Miss Buss’ endeavours to respond to
even the most unreasonable of demands. When she
met me at Ben Rhydding soon afterwards, she was
just as sweet and bright as ever, and her nerves
rapidly recovered tone again. This power of
recuperation after even the severest strain was
always remarkable, even to the very last. We had a
striking proof of it in the spring of 1893, when Miss
Buss joined my sister and me at Bordighera. We had
tried to get her to take the complete rest of a whole
winter abroad after her illness in the autumn before,
holding out the attractions of Florence, Siena, and
the Italian lakes. Every one wanted her to give up
work for a time, and take the chance of real
recovery. Our efforts were all wasted, and all she
would do was to come, with her cousin, Miss Mary
Buss, and a friend, late in the spring, stopping at
various points in the Riviera on the way. She was far
from well on her arrival, but a drive to San Remo in
an open carriage on a windy day gave her a chill,
followed by the inevitable attack of influenza. There
was also a passing giddiness which gave us anxiety.
She was certainly very ill for five days, with a
threatening of pneumonia. But, thanks to her power
of sleeping day and night, the attack passed off as
rapidly as it had come on, when nothing we could
say could persuade her that there had been ground
for alarm; an opinion she maintained in the face of
the most authoritative medical support of our view.
On the Sunday she had certainly been very ill, but on
Tuesday she would have been downstairs if we had
not made too strong a protest. On Thursday,
however, she insisted on starting for England, and
accomplished the journey to London without a break,
and apparently with no ill consequences.
She had already suffered from frequent attacks of
influenza of a more or less serious character, leaving
behind them more and more weakness. The first
attack dated from the winter of 1889–90, when we
were all in Rome together. I had suffered from what
seemed a sudden sharp cold, but was nearly well
when Miss Buss and her party arrived in Rome on
Christmas Eve. Christmas Day was very wet, and as
my room was large and airy all assembled there for
afternoon tea and talk, Miss Buss being full of fun
and interest. But after a few days she and several
others developed the same kind of cold, which, even
then, we never identified with the mysterious disease
of which every one heard so much that year. But for
us both it proved the beginning of a series of attacks
extending through the next four years. More than
once when she was at the worst, I was too ill even to
be told of it till the danger had passed. This was the
case in the autumn of 1893, and I had been
suffering during the summer, and able to see her
only when she came to visit me.
It was during this summer that she finally moved
from Myra Lodge to No. 87, next door, leaving the
boarders with Miss Edwards. The door of
communication was still left, that Miss Buss might
see her friends and the girls when she felt able. She
had her own companion, Miss Newman, and, later,
Miss Millner; but Miss Edwards, having been so many
years with her, still went often to see her. There
seemed every prospect of years of rest and ease,
amid a circle which could profit by her experience
and wisdom.
There were all the inevitable delays, in getting into
the new house, even though the workmen worked
with all their hearts for an employer who took very
special care of their creature comforts, and made
them wish “for more like her.” She was not
accustomed to summer in London, and the
consequence of it all was the very serious attack,
already mentioned, in the autumn. She recovered,
however, with something of the rapidity of the
experience in the spring, and was able to go to
Bournemouth, and afterwards to spend Christmas at
her cottage at Epping.
 When my sister and I returned from Italy, in May,
1894, we were very much grieved to see the change
in our friend. She looked many years older, and was
quite unfit for any sort of exertion. It was surprising
how easily she accepted the changed conditions,
and, after her life of so much activity, was quite
content to be amused, finding special pleasure in
Miss Millner’s lovely little Persian kitten. It was very
touching to see her intense amusement in her
subjection to her new medical attendant, Dr.
Cobbett, the successor to her old friend Dr. Evershed.
She even seemed to find a lively satisfaction in the
discovery of a will which could dominate her own.
There was one bright spot in this summer, in a visit
to “The Haven,” near Hythe, the pleasant home of
her friend Mrs. Pierson, from which she returned so
well that she went to the Norfolk coast with Miss
Millner and Mr. and Mrs. W. K. Hill. But the weather
was cold, and Mr. and Mrs. Septimus Buss, who
joined her at Overstrand, were thankful to get her
safely home again.
The change in our dear friend, with the manifest
certainty that she must soon retire from her work,
had made me look out her old letters, and begin to
arrange the material she had prepared for the long-
talked-of story of the school, which I wished to have
ready when the day of retirement should come. On
my last visit to her, early in November, she was so
much stronger that she talked in quite the old way,
telling me that she intended to amuse herself by
dictating her reminiscences to a shorthand writer. I
then told her what I had been doing, and she
became quite eager that we should do it together.
On November 12th I had a note fixing the next day
for the first of these meetings. I was unfortunately
prevented from going, thus losing that last precious
evening of her active life—a lasting regret.
Early the next day the fatal illness began with an
attack of unconsciousness. In a letter from the Rev.
Alfred J. Buss, he says—

“Though my sister had been in ill health for a long time,

she had rallied so often that much hope still remained. She
had been at the school several times during last term, and
attended a meeting of ‘old pupils.’ This last may have been
too much for her. She had an attack from which she was
unable to rally. There had been a consultation a few days
before, and the medical men saw no reason why she should
not then be better—and allowed me to inform the governors
so—though she would still be liable to relapses. So that the
end came unexpectedly.”

She had recovered from several similar attacks,

and had latterly seemed so much stronger that there
had been every reason for hope during the periods
of consciousness that came from time to time,
although a new symptom had appeared in the
extreme restlessness that alternated with the
lethargy.
For six weeks hope came and went, everything
being done that love could devise or devotion carry
out. In addition to the two constant companions,
there were two trained nurses: and the dear patient,
in the quiet intervals, was her sweetest self; so
careful about giving trouble, and so courteous in her
acknowledgment of service rendered, so grieved that
the nurses should be kept up at night, and so
anxious that Miss Millner and Miss Edwards should
know how much she felt their kind attention.
Miss Edwards gives some interesting details of
these last months after the return home from this
last holiday, when, after a few weeks of care and
nursing, she had seemed better than at any time
during the year:—

“Three weeks of peaceful, quiet enjoyment followed this

illness, during which Miss Buss received many of her friends
at her own house, and was further made happy by a visit
from her old and intimate friend, Mrs. Hodgson, who has
since written: ‘I am very thankful that I had such a sweet,
happy time with my friend before the last illness came, and
when she could in a measure enjoy life.’
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebookluna.com