3rd International Conference on

PROTECTION OF HISTORICAL CONSTRUCTIONS

Lisbon, Portugal, 12 – 15 July, 2017

HISTORICAL EARTHQUAKES AND HISTORICAL BUILDING

ARCHETYPES FOR DYNAMIC STRUCTURAL ANALYSIS USING
HISTORICAL SURVEYS

Eduardo Charters Morais*, László G. Vigh* and János Krähling**

* Structural Engineering Department, Budapest University of Technology and Economics, Budapest, Hungary
e-mails: eduardo.charters@epito.bme.hu, vigh.l.gergely@epito.bme.hu
**Department of History of Architecture and Monument Preservation, Budapest University of Technology and
Economics, Budapest, Hungary
e-mail: krahling@eptort.bme.hu

Keywords: Historical survey; Historical earthquake; Building Archetypes; Masonry buildings; In-plan
Indices; Dynamic Structural Analysis;

Abstract. The development of structural archetypes for dynamic structural analysis is a current trend
in seismic risk assessment and damage scenarios for urban environments. These processes usually rely
on probabilistic seismic hazard analysis (PSHA) to select the ground motions for incremental dynamic
or multiple stripes analysis in order to attain building specific fragility functions. A method for the
estimation of the magnitude of historical earthquakes uses fragility data together with the framework of
PSHA, combining seismic damage records of historical structures with the knowledge of the probable
seismic sources. Hence, the estimation of the magnitude of the 1763 Komárom earthquake (West of
historical Hungary) requires the development of archetypes of the affected buildings. To do so, both
specific bibliography and the historical survey of the city of Tata are used, the historical buildings are
categorized and qualitatively analysed in order to draw the 18th century structures. A parametric and
index based numerical framework is developed, providing both the bounds for archetype generation and
the basis for a simplified vulnerability assessment. Four archetypes are presented and analysed.

1 INTRODUCTION
The great majority of the seismic events occurred long before the advent of modern and
instrumental seismology (20th century). Nonetheless, seismologists usually engage in the study
and estimation of the magnitude of historical earthquakes using historical (written sources and
depictions) and geological data of their macroseismic effects [1,2]. The study of these events is
relevant to understand the regional seismicity and improve the safety on new structures. The
most usual methods for the estimation of the magnitude use an intensity scale, as the Modified
Mercalli Intensity (MMS), to achieve intensity point(s) that help to determine the epicentral
intensity, then they rely on empirical relationships relating the magnitude with the epicentral
intensity (or radius) to achieve magnitude estimates [3]. Although, these methods greatly
simplify the behaviour of real structures affected by the earthquake to achieve those estimates.
A different approach estimates the magnitude considering the nonlinear seismic behaviour
of structures and uses the Probabilistic Seismic Hazard Analysis (PSHA) framework to
combine historical seismic records with fragility functions – using pushover based Capacity
Spectrum Method (CSM) [4] or Dynamic Structural Analysis (DSA) [5,6] – and data on the
regional seismic sources [7]. Although, the application of the method raises questions about the
appropriateness of the structures selected for analysis. A possible path is to develop historical
buildings’ structural archetypes, as those in FEMA HAZUS earthquake model [8], representing
the features of the buildings in the region affected by the historical earthquake. Afterwards,
 E. C. Morais et al.

nonlinear CSM or DSA are used to generate fragility functions representing the seismic risk
and damage scenarios for urban environment [9,10]. An earlier study introduced a framework
for a flexible, comprehensive and quantitative approach for the generation of structural
archetypes for DSA that uses historical building surveys [11], focussing on the 1763 Komárom
earthquake (18thc historical Hungary), but it has not been fully applied. Furthermore, this study
came to question an earlier expert based historical archetype, for which Incremental Dynamic
Analysis (IDA) [12] and preliminary magnitude values were estimated [13].
The present paper focusses on the generation of historical buildings’ archetypes to estimate
the magnitude of the Komárom 1763 earthquake, using the historical survey of the city of Tata
[14] and the archetype generation methodology introduced in [11]. The great majority of the
buildings were composed by complex masonry elements leading to difficult uncertainties. To
carry out this study and clarify its scope and conditions, the magnitude estimation method and
fragility functions generation process are briefly explained, and the historical sources for the
1763 Komárom earthquake are presented and discussed. In the third section the results of the
study of the historical survey of the city of Tata are presented regarding. The archetypes are
presented and analysed in the last section. Conclusions and observation remarks are presented.

2 METHODOLOGY

2.1 Magnitude estimation

The magnitude estimation method in (fig. 1) [7] relies on fragility functions of the type
P(dmi|im,m,r), relating an intensity measure (IM – which can be PGA, PGV, Sa(Ti), etc.),
magnitude, M, and distance, R, to a probability of a certain damage state (DM). These can be
achieved either by empirical, or analytical or hybrid methods [15]. Afterwards, the fragility
curves are combined with historical structural damage records associated with the event E in
study. This early step in the method, that culminates in probability of the event E as P(E|im,m,r),
assumes the correspondence between the analysed structure(s), or structure type(s), with the
number and types of damages occurred due to the event. Afterwards, the total probability
theorem enables to integrate the probability P(E|im,m,r) and the attenuation relationship (or
distribution of IM) and prior distribution of R, respectively fIM|M,R and fR, in order to calculate
the probability P(E|m). In the end, the Bayes’ theorem is aplpied to achieve the posterior
distribution of magnitude fM|E using a prior distribution of magnitude fM.

Figure 1: Magnitude estimation methodology, based on [7].

Despite earlier studies regarding this method, the assumption that states the link between the
structures under analysis and the occurred damages has been questioned in [12,13] where a 18th

2
 E. C. Morais et al.

century structure of a peasant house was assumed for DSA. This led to the development of a
methodology to generate historical building archetypes from historical building surveys [11].
The archetypes, ‘structures that fairly represent the range of configurations and properties of
the building group of interest’ [8], should represent as much as possible the building-damage
relation, but also the historical sources and literature describing those buildings.

2.2 Historical and contemporary sources

The 1763 seismic event, with epicentre near the city of Komárom, was one of the strongest
and most destructive in Hungarian history, estimated with a magnitude in between 5.7 and 6.5.
Therefore, early and contemporary sources figure between the evaluation of the seismic damage
[3,16] and of the understanding of the seismicity of the region [1]. Sources refer to 91% of
damaged buildings in the city of Komárom, in different degrees: 279 completely destroyed, 353
partially collapsed, 213 needed expensive repairs and 219 cheap repairs [2]. Other sources refer
to the most affected building types as the tall and stiff public buildings and manors, rather than
the one-storey and more flexible commoners’ rammed earth and adobe buildings [3]. Several
depictions (fig. 2) enable the identification of the most probable failure modes, showing show
low-rise (one, two storey) masonry baroque buildings with cracks on their walls’ surface,
collapsed towers and gables, both vertical and diagonal cracks in walls, and roof displacements.
Monumental buildings were also affected, although they fall out of present scope.

Figure 2: Karl Friedel (left) and anonymous’ (right) depictions of the damage caused by 1763
Komárom earthquake.

There is actually no broad accessible historical building survey of the city of Komárom.
Although, the city of Tata is close to Komárom, it is broadly surveyed and holds similarities.
Therefore, its historical survey [14] may be used to source the archetype generation besides
historical architecture literature [17,18,19] and the depictions of Komárom (fig. 2) as in [11].

2.3 Baroque townscape of Tata

The medieval town of Tata firstly flourished around the squared form castle in the 15th
century under the rule of King Sigismund von Luxemburg and Mathias. The royal castle (later
incremented with bastions) played later an important role in the wars against the Turkish army,
as an important preceding fortification before the stronghold of Komárom/Komarno. It was
several times occupied periodically by the Turks.
The historic townscape of Tata developed after the Ottoman rule – after 1686 – transforming
a ruined medieval town fabric into a baroque townscape. Count Joseph Eszterházy, a Lord Chief
Justice in Hungarian royal administration purchased the town in 1727 holding it by the family
for two more centuries. By the time of the Komárom earthquake the most important public
building of the town, the parish church was already under construction from 1751, built
according to the plans of Franz Anton Pilgram and executed by the famous local master builder

3
 E. C. Morais et al.

Jakab Fellner [18]. Most of the 18thc Tata baroque townscape may be associated to the work of
Jakab Fellner who designed under the landowning of the Eszterházy for different societal
groups: houses for ecclesiastic purposes like monasteries, parish houses and charity institutions,
inns, mills, manufactories for porcelain, leather production etc., buildings for agricultural
purposes. He provided plans for a new pompous palace to be built on the plot of the ruinous
castle, idea not realized due to the unrealistic demolition works. The “Small Castle” built for
the steward-of- estate Ferenc Balogh preceded the plans of the bigger Castle built from 1765
that made – with the parish church project – fame for Fellner in the Eszterházy Estate and
beyond. His architecture contributed to the balanced and qualitative shape of the baroque
market-town, although most of his drawings are currently lost.
The historic development of Tata can then be characterized as an organic growth of a town
consisting of a medieval historic centre close to the fortification that had been extended with a
new street system adapted to the geographical environment.

2.4 Archetypes generation

The survey is composed by plans of historical buildings from Tata, some of them have been
strongly modified and other do not fit the interest period (18thc). Even after careful selection it
is not obvious how to depart from a historical survey and create perfectly measured structural
archetypes. One answer to the question was addressed in [11] resulting in a method composed
by two main analysis movements: one qualitative and one quantitative (fig. 3). Once selected
and interpreted, the buildings are categorized according to sources, the plans are measured and
the historical parts and main structural systems are identified, as well as the typical sections and
plans. Walls, with openings or not, arches, vaults figure among these, although the walls are
taken as the main earthquake resisting system matching both the depictions (fig. 2) and
descriptions (may not be directly related to the event). Afterwards, the basic and output/control
variables are defined and the measurements are carried out and gathered in order to calculate
the three in-plan indices [20,21] and define upper and lower bounds. The archetypes are then
composed by framing the layouts from the qualitative analysis in the bounds obtained from the
quantitative analysis of the historical survey. The methodology is here synthesized:

Figure 3: Description of the methodology for archetype generation, based on [11].

The following sections present a description of the categorization matching in the qualitative
analysis and the results of quantitative analysis for the historical survey of Tata, for each
described category and for the relevant variables to be used as bounds for archetypes generation.
These processes are more generally described elsewhere [11].

4
 E. C. Morais et al.

3 HISTORICAL BUILDING SURVEY

In order to target the 1763 earthquake, the west of the Hungarian 18th century kingdom was
set as interest region and period, but still there exist distinct buildings belonging to different
categories. Therefore, the building plans were gathered from the historical survey of the city of
Tata and literature on monuments of the city [14,18], being double checked using historical
monuments using historical monuments literature [17,18,19] in order to match the period
criteria and attain the historical intervention descriptions. Afterwards, the buildings were
categorized as to their function and analysed in order to draw the 18thc structure from the
existing building before the measurements. The description of the categories and the
quantitative analysis of both basic variables and output variables, the in-pan indices [20,21],
are presented in this section.

3.1 Description and matching of category frameworks

The identification of the buildings’ categories, partially carried out in [11], selected and
organized 34 buildings out of 122, concluded the existence of 7 categories: residential or
dwellings (52.9%), manors (11.8%), commercial (2.9%), public buildings (2.9%), industrial
buildings (11.8%), religious buildings (14.7%) and military buildings (2.9%). The same study
enabled the identification of sub-categories within the main category framework as some of the
manor houses (5.9%) may be placed in relation to residential buildings and the religious
buildings are divided in three sub-categories: monasteries (5.9%), churches (5.9%) and chapels
(2.9%). Although, this category division brings conflict with the approach of the Consilium
Locumtenentiale, in the 18thc, regarding the damage/repair costs [16] (and befitting the social
division of society), the two categories are not coincident. As a result, not all the dwellings were
identified as taxpayers’ houses as some of them were noblemen houses. Therefore, the 7 sub-
categories regrouped into four basic categories associated to the damage reports [16]. While
religious buildings, as churches, parochial houses and chapels compose the first category
(17.6%), castles, palaces and manors compose the second (11.8%), public buildings, inns, mills
and kilns the third (26.5%), and taxpayers’ houses the fourth (44.1%), (fig. 4, right).

Figure 4: Buildings’ categories charts, based on [11] (left), and according to [16] (right).

The great majority of the surveyed buildings (91.2%) is low-rise – with one or two storeys
high – with or without basement. In this matter the Capuchin and Parish churches, as well as
the city castle are exceptions, as they rise to greater heights. From the total number of buildings
11.8% possess a basement. One reason may be the proximity of the lake, which raises the water
level. From the total number of taxpayer houses 33% are one-storey and the remaining 66%
two-storey buildings, and only one building, which is, one-storey high possess a basement. This
qualitative analysis enables the identification of typical section layouts and plans.
Table 1: Buildings by category, and plan and section types.
ID 1 2 3 4 Plan Layout ID 1 2 3 4 Plan Layout ID 1 2 3 4 Plan Layout

5
 E. C. Morais et al.

001 × I 1 014 × I or U 3 027 × L 8

002 × I 1 015 × U 1 028 × I or L 9
003 × I 8 or 12 016 × L 3 029A × I 11
004 × I 9 017 × I or U 3 029B × U 1 or 2
005 × I 11 018 × L 5 or 10 030 × U or □ 10
006 × I 3 or 8 019 × I 1 031 × I 7
007 × I or L 7 020 × I - 032 × U -
008 × I 9 021 × U or □ 1 or 5 033 × I or L 7
009 × I 1 022 × I - 034A × I 10
010 × I 1 023 × I 1 034B × I 4
011 × I - 024 × I 3 034C × I 1
012 × U 5 025 × I 1 or 5
013 × L 7 026 × U 7

A synthesis of this analysis is shown in figure 4, while an illustration is available in section

4. This table ignores the aspects of churches, chapel and castle generalization (because of the
small sample). Furthermore, the ground plan interpretation is not univocal and the plan
arrangement may have base for speculation. Therefore, different hypotheses are taken here as
valid and incorporated for forthcoming interpretation and archetype generation.

3.2 Definition and analysis of variables

3.2.1 Definition of basic and output variables
The adopted in-plan indices [20,21] have been proposed in order to provide a simplified pre-
evaluation of the seismic vulnerability of historical masonry buildings, although, in the present
case it is argued that they can also be used as numerical criteria to generate archetypes befitting
the CSM or DSA based fragility curves generation. The in-plan area ratio, γ1,i, the in-plan area
to weight ratio, γ2,i, and the base shear ratio, γ3,i, can be calculated as follows:

Awi A F
 1,i   1.0 ,  2,i  wi  1.2 &  3,i  Rm,i  1.0 (1)
S G FEm

Where Aw,i is the in-plan wall area in i (x or y direction), S is the total wall area, G is the total
weight of the building, FRm is the median shear strength of the structure and FEm is the median
seismic action force. While FRm= Σ𝐴𝑤,×[fvk0+0.4(γm×hm)], FEm=β×Gm, being β a coefficient
related to the ground acceleration (it may be assumed as the PGA or the SAmax). The variables
have been changed from previous applications in order to address median instead of design
conditions (FRd and FEd). Other basic variables were also assumed as the specific weight,
γm=18kN/m3, cohesion, fvk0=0.10MPa, friction angle, φ=22º, although, these assumptions –
following [21] – require a more detailed evaluation of strength parameters, although they are
arguably appropriate for burnt clay masonry. This set is here taken as case study 1 for tax-
payers’ houses. The case study 2 is set for adobe structures, which are smaller, lighter and more
flexible but also weaker, with γm=16kN/m3, fvk0=0.044MPa and φ=34º [23]. Furthermore, in
order to calibrate the coefficient β, ground motion prediction equation (GMPE) in [22] was used
to calculate the acceleration response of the structure. As previously mentioned both
PGA≈0.25g or averaged SAmax(0.20<T<0.50)≈0.52g may be used, although the later seems to
be conceptually more appropriate for rigid low-rise masonry structures instead of PGA which
is equivalent to the spectral acceleration at nearly zero period SA(T≈0).
3.2.2 Distribution of basic variables
As mentioned in [11] the basic measured variables are the height, h, one thickness, teq
(equivalent), two span lengths, Lmin and Lmax (minimum and maximum) and three areas Aw,x,

6
 E. C. Morais et al.

Aw,y and S. One important question is how to condition the variables to build an archetype. An
a priori idea is to hold the minimum and maximum measured values and use them as upper and
lower bounds, but it leaves the method sensitive to outliers. Another strategy is to identify
outliers and provide a statistical model to fit the data, outliers would still have a residual
probability of being observed but the bounds would be defined in the interval of median more
or less a standard deviation (µ±σ). One example illustrates the distribution of height, for which
normal and lognormal fits are applied: two-storeyed dwellings (or tax-payers’ houses – fig. 5,
left) and all the surveyed buildings (fig. 5, right). This exercise shows, in the first case, that the
normal distribution fits better than the lognormal distribution (2.99±1.18m). Although, if we
unite all the categories in a single set, the lognormal distribution (3.27±1.27m) is better.

Figure 5: Normal (in blue) and lognormal (in red) fits over the heights histograms of dwellings (left)
and for all categories (right).

Due to the small difference between the distributions, only the arithmetic means and standard
deviations are presented here. The results (table 2) show a general but considerable discrepancy
in the geometric parameters between categories, but a consistent decreasing from the first to the
forth category. This is consistent with the social division of society, in the sense that richer
classes possess bigger buildings, and that religious buildings generally prevail in terms of
proportion over the noblemen houses, with higher heights and average wall areas.
Table 2: Statistical study of the input variables assuming a normal distribution, µN(σN).
h [m] teq [m] Lmin [m] Lmax [m] Aw,x [m2] Aw,y [m2] S [m2]
4.16(1.61) 1.00(0.37) 2.89(1.46) 7.10(3.07) 140.33(152.28) 133.61(156.37) 244.65(294.24)
3.63(0.65) 0.96(0.38) 2.81(2.13) 10.50(3.99) 105.18(81.38) 88.97(60.29) 161.80(115.16)
3.20(0.57) 0.73(0.19) 1.87(0.34) 5.95(0.98) 54.48(30.89) 46.92(28.35) 89.97(53.79)
3.02(0.45) 0.55(0.08) 2.35(1.18) 5.45(0.77) 31.25(25.25) 23.93(22.22) 49.68(43.27)
3.36(0.89) 0.75(0.31 2.42(1.36) 6.86(2.86) 69.27(86.1) 60.85(83.03) 113.82(154.25)

3.2.3 Distribution of output variables

Despite the total wall heights, lengths and thicknesses, and in-plane areas, specific for every
building, to build archetypes, with average or sampled geometrical parameters requires the
rationalization of these parameters. This is the main reason why the study of the output variables
focusses on the three in-plan indices (equation 1). In order to illustrate the use of γ indices, the
exercise is carried out for the fourth category (taxpayers’ houses), implementing the two case
studies mentioned in 3.2.1 (for clay and adobe masonry, respectively). The results can be seen
in table 3. The γ1 index, being a pure geometrical ratio, has the same value for the two cases.
Table 3: In-plan indices for the fourth category: tax-payers’ houses (1 and 2 storeys). The green and
orange colours relate, by satisfying or not, respectively, the conditions in equations (1).
Case 1 Case 2

7
 E. C. Morais et al.

ID γ1,x γ1,y γ2,x γ2,y γ3,x γ3,y γ2,x γ2,y γ3,x γ3,y
001 0.58 0.49 9.51 8.13 3.30 2.82 10.70 9.14 2.12 1.81
002 0.60 0.52 7.81 6.81 2.43 2.12 8.78 7.66 1.87 1.63
006 0.62 0.52 9.10 7.73 2.63 2.24 10.24 8.70 2.00 1.70
009 0.61 0.53 9.27 8.09 4.40 3.84 10.43 9.10 3.02 2.64
010 0.70 0.39 18.99 10.56 5.81 3.23 21.37 11.88 3.99 2.22
1 STOREY

0.69 0.50 14.51 10.51 4.05 2.93 16.33 11.83 2.83 2.05
015 0.74 0.36 14.94 7.28 5.10 2.49 16.81 8.19 3.48 1.70
019 0.73 0.34 13.76 6.44 3.68 1.72 15.48 7.24 2.71 1.27
021 0.59 0.51 9.96 8.56 0.82 0.70 11.20 9.63 0.93 0.80
023 0.89 0.85 8.93 8.59 3.64 3.50 10.05 9.66 2.55 2.45
024 0.69 0.38 11.26 6.27 1.83 1.02 12.66 7.05 1.61 0.89
034B 0.67 0.44 10.25 6.85 3.29 2.20 11.53 7.70 2.42 1.62
034C 0.68 0.41 12.85 7.72 5.06 3.04 14.46 8.69 3.46 2.08
013 0.69 0.41 12.34 7.36 1.97 1.17 13.89 8.28 1.68 1.00
0.58 0.50 10.04 8.60 2.90 2.48 11.30 9.67 2.14 1.83
016 0.47 0.64 7.86 10.72 1.58 2.15 8.85 12.06 1.29 1.76
2 STOREYS

0.59 0.49 11.50 9.55 4.73 3.93 12.93 10.74 3.23 2.68
026 0.65 0.44 8.85 5.93 1.28 0.86 9.95 6.67 1.25 0.84
0.72 0.34 10.29 4.85 2.40 1.13 11.57 5.46 1.96 0.93
027 0.56 0.55 8.75 8.61 1.13 1.12 9.84 9.69 1.09 1.07
0.63 0.48 9.88 7.48 2.13 1.61 11.12 8.41 1.72 1.30
034A 0.60 0.51 9.19 7.91 1.50 1.29 10.33 8.90 1.33 1.15
0.65 0.48 11.04 8.12 2.84 2.09 12.42 9.13 2.14 1.57

The results show that despite the reduction in the specific weight of the structures (increasing
γ2 values) and the increase in the friction angle, from case 1 to case 2, this was not enough to
cover the general reduction in cohesion, leading to lower γ3 average values (table 2, fourth row),
and one more structure with insufficient strength (ID024, table 3, in orange). Due to the
simplified analysis provided by the indices and comparative nature of the study, the real
dynamic behaviour of the structure is not included in the analysis, as well as the nonlinear
relation between strength and deformation, actual material parameters, variability of the ground
acceleration (approx. 34%), and epistemic uncertainties regarding the building survey.
Nonetheless, both the case study and the average values in tables 3 and 4, respectively, may
provide a rough dimension of the 1763 event and arguable validity of the selected buildings.
Table 4: Statistical study of the output variables assuming a normal distribution fit, µN(σN).
Case 1 Case 2
γ1,x γ1,y γ2,x γ2,y γ3,x γ3,y γ2,x γ2,y γ3,x γ3,y
0.62(0.15) 0.55(0.12) 7.13(3.11) 6.83(4.45) 1.01(0.72) 0.89(0.61) - - - -
0.62(0.09) 0.56(0.07) 8.28(1.78) 7.49(1.55) 1.02(0.50) 0.97(0.58) - - - -
0.62(0.04) 0.53(0.05) 10.09(1.61) 8.59(1.13) 1.53(0.67) 1.31(0.62) - - - -
0.65(0.08) 0.48(0.11) 10.91(2.66) 7.94(1.47) 2.98(1.40) 2.16(0.96) 13.01(3.26) 9.46(1.79) 2.30(0.89) 1.67(0.60)
0.63(0.09) 0.52(0.10) 9.55(2.74) 7.81(2.23) 1.92(1.33) 1.51(0.93) 10.10(3.19) 8.21(2.36) 1.58(0.84) 1.27(0.62)

One conclusion that may be drawn from this study is that the index values associated with
tax-payers’ houses is associated with higher base shear ratios then most of the religious,
noblemen houses and industrial buildings. This early result show not only a deficiency in the
strength FRm in face of the seismic action FEm of the later categories of buildings (1, 2 and 3) –
in comparison with category 4 – but also verifies the historical damage descriptions in this
particular claim. In total, 14 out of 34 buildings reveal a deficiency in γ3 – despite G being
underestimated in this study – but only one in γ2. Although, the study cases 1 and 2 suggests
that the reason under the resilience of adobe masonry houses would have been more related to

8
 E. C. Morais et al.

having a lower weight (due to geometry and low specific weight) rather than strength properties.
Another explanation is the higher flexibility of adobe, which cannot be judged here.

4 ARCHETYPE GENERATION
Once the bases for both the qualitative and the quantitative framework are assembled, the
generation of the archetypes may be carried out. The strategy resides in the use of average or
sampled values of basic parameters and gather them using both a plan and a building section
associated to the survey buildings. Afterwards the bounds of the in-plan indices are used to
validate the archetypes. The process is here illustrated, a previously used archetype [11,12] is
evaluated and new archetypes are presented.

4.1 The archetypical layouts

Despite the categorization in 3.1 and table 1 synthesis the typical plans and building section
were not illustrated. Therefore, the typical building section layouts (numbered in table 1) find
their reference in figure 6. The typification neglects the existence of basements, which has small
significance in the surveyed buildings (11.7%), and no apparent effect in the in-plan strength
of the walls. Simplifications went further as churches and chapels layouts are also neglected
because of insufficient sample and relatively small significance in the damage descriptions
related to the global behaviour of the affected buildings. This argument may be seen with more
detail in [11]. Despite ambiguity in the interpretation, hypothesis for the layout were included
as additional type cases (i.e. table 1, ID003, type ‘8 or 12’).

Cat. 1 1 - - - 2 -
Cat. 2 1 1 - - 1 -
Cat. 3 - - 2 - - 1
Cat. 4 9 - 2 1 1 -

Cat. 1 - - - 1 - -
Cat. 2 - - 1 1 1 -
Cat. 3 3 - 2 - 1 1
Cat. 4 2 2 - 1 - -
Figure 6: Archetypical layouts, based on [11,13].

As can be seen, the buildings are generally defined by basic cells, composed by walls and
openings, either vaulted or not. A small number of buildings possess columns supporting arches
and vaults (i.e. layout 6, cat. 3). The building sections are typically one or two storeyed and
composed by one or two spaces, or depths, with a corridor or not. The study is not conclusive
although, in general, the layout 1 and not-vaulted spaces are dominant for category 4: 50% for
the former and 83% for the later. In respect to the plan shape, and according to table 1, the ‘I’
shaped plan is dominant in all categories (56.8%), followed by ‘U’, ‘L’ and ‘□’ shapes (20.5%,
18.2% and 4.5% respectively). The dominance of an elongated shape is also coherent with the
small variability of γ1 values (table 4). While the ‘□’ shapes may be considered accidental for
category 4, it may be not considered for category 1, as several other monasteries follow that

9
 E. C. Morais et al.

configuration. ‘L’ shapes are common in historical Tata and in agreement with the historical
map presented in [11]. Another most common feature is the gabled roof.

4.2 Archetypes and parametrical adaptation

In a previous work [11] an archetype of a peasant house (figure 7, left) was evaluated and
compared using the archetype generation method. It resulted that γ3 were not appropriate (table
5, A0). Although, changing it to adobe and adapting the input parameters to average values for
category 4 ratios changed to fall into the intervals of table 4 (table 5, A1). Other archetypes were
composed following the same approach (figures 7 and 8), for which the input parameters (table
5) were selected to respect the control parameters (γ) in table 4.

Figure 7: Peasant house (left), multi-category (centre) and mill (right) archetypes.

These archetypes result from additional interpretation considering both socio-economic and
geographical conditions together with the categorization and literature. This endorsed not only
the importance of assembling an archetype of a peasant house, which is highly representative
of the built environment, but also of a mill (A6, fig 7, right) mainly because of the existence of
the lake and canals which source a considerable number of mills. It is also considered a multi-
category archetype (fig. 7, centre), which is adaptable to become either a city house (A2), or an
inn (A3), or a noblemen house (A4) or a parish house (A5), depending on the adapted parameters.

10
 E. C. Morais et al.

Figure 8: Noblemen house archetype.

Table 5: Input and control parameters (γ) for the building archetypes (fig. 7 & 8).
L1 L2 L3 L4 L5 L6 L7 t1 t2 h1 h2 h3 γ1,x γ1,y γ2,x γ2,y γ3,x γ3,y
A0 4.70 2.80 3.80 4.70 - - - 0.50 - 2.60 2.00 - 0.64 0.48 12.23 9.14 5.8 4.34
A1 4.70 2.80 3.80 4.70 - - - 0.55 - 3.00 2.50 - 0.67 0.46 12.31 8.52 2.97 2.06
A2 4.50 4.50 3.50 4.50 4.50 5.00 2.50 0.55 0.55 3.00 2.50 - 0.63 0.45 10.83 7.69 2.52 1.79
A3 4.50 4.50 3.50 4.50 4.50 5.40 2.50 0.73 0.73 3.20 3.00 - 0.62 0.47 9.9 7.46 1.83 1.38
A4 4.80 4.80 3.80 4.80 4.80 6.20 3.25 0.96 0.96 3.63 3.20 - 0.61 0.49 8.51 6.79 1.24 0.99
A5 4.80 4.80 3.80 4.80 4.80 6.20 3.25 0.96 0.96 4.16 3.50 - 0.62 0.49 7.43 5.92 1.11 0.88
A6 5.95 3.25 4.45 4.45 - - - 0.73 - 3.20 3.20 6.40 0.66 0.44 10.44 6.93 1.60 1.06
A7 5.49 4.57 3.66 4.57 5.49 4.57 4.57 0.96 0.96 3.63 3.63 3.63 0.63 0.56 8.12 7.19 0.92 0.82

Other building archetypes such as castles, churches and monasteries are not here presented.
They require a broader study and pose small representability of the damage caused by the 1763
seismic event. Another remark is that the nobleman house archetype (A7, fig. 8) is identical to
the Eszterházy castle in Tata, which is a two depth baroque building with five rows, although,
it may be adapted for smaller low-to-middle class noblemen, which are typically two depth
buildings with three rows.

5 CONCLUSIONS
This paper addressed the development of structural archetypes for historical buildings related
to the damage caused by the 1763 Komárom earthquake. A method, suggested in earlier studies
enabled both the identification and categorization of the typical layouts and the development of
a numerical framework with bounds for archetype generation. Four new archetypes are here
presented and discussed, focussing on the four main categories. It can be verified that the
numerical framework works comparatively to generate valid archetypes as far as the estimation
of the parameters is standard for the surveyed buildings, although, it only works as a simplified
vulnerability assessment (of both buildings and archetypes) if parameters as the total weight,
G, or γ indices are thoroughly estimated. The framework may always be improved in this
regards. Furthermore, out-plan indices may, as well as other relevant parameters may be added
to the numerical framework. The issue of the roof systems was not discussed despite the
varieties in the late baroque period in Hungary, which is covered by literature. Although, it was
demonstrated that it is possible to merge quantitative and qualitative analysis to assemble
historical buildings archetypes in order to address the damage caused by the 1763 Komárom
seismic event.

AKNOWLEDGEMENTS
The Tata historical survey (1993) was supervised by Professor Istvánfi Gyula and Professor
Mezős Tamás with the associated staff, for the subject of ‘Monument Survey’ created by
Professor Hajnóczi Gyula (late 1970’s), in the ‘Institute of History and Theory of Architecture’
(Budapest University of Technology and Economics. Proper acknowledgements to CAPES for
allowing the present research (Process No. 9178-13-9). This paper was also supported by the
János Bolyai Research Scholarship of the Hungarian Academy of Sciences.

REFERENCES

[1] Zsíros T., “Seismicity of Komárom-Mór Area”, Acta Geod. et Geoph. Hung, 39(1), 121–131, 2003.
[2] Varga P., Szeidovitz G. and Gutdeutsch R., “Isoseismical map and tectonical position of the
Komárom earthquake of 1763”, Acta Geod. Geoph, 36(1), 97–108, 2000.

11
 E. C. Morais et al.

[3] Grossinger J., Dissertatio de terrae motibus regni Hungariae, Győr, 1783.
[4] Freeman S.A., “The capacity spectrum method as a tool for seismic design”, Proceedings of the
11th European Conference on Earthquake Engineering, Paris, France, (08/09), 625, 1998.
[5] Vamvatsikos D. and Cornell C.A., “Incremental dynamic analysis”, Earthquake Engineering &
Structural Dynamics, 31(3), 491-514, 2002.
[6] Jalayer F. and Cornell C.A., “Alternative nonlinear demand estimation methods for probability-
based seismic assessments”, Earthquake Eng. & Structural Dynamics, 38(8), 951-972, 2008.
[7] Ryu H., Kim J.K. and Baker J.W., “A probabilistic method for the magnitude estimation of a
historical damaging earthquake using structural fragility functions”, Bulletin of the Seismological
Society of America, 99(2A), 520-537, 2009.
[8] HAZUS-MH MR-1 Technical Manual, FEMA, Washington D.C, 2003.
[9] Kegyes-Brassai O.K., “Earthquake hazard analysis and building vulnerability assessment to
determine the seismic risk of existing buildings in an urban area”, PhD dissertation, Széchenyi
István University, Győr, Hungary, 2014.
[10] Lamego, P.R., “Seismic strengthening of residential buildings: risk analysis and mitigation” (in
Portuguese), PhD thesis, University of Minho, Guimarães, Portugal, 2014.
[11] Morais E.C., Vigh L.G. and Krähling J., “Development of historical building archetypes for seismic
performance assessment”, Pollack Periodica, 2017. (accepted)
[12] Morais E.C., Vigh L.G. and Krähling J., “Fragility estimation and comparison using IDA and
simplified macro-modeling of in-plane shear in old masonry walls”, Springer Proceedings of
Mathematics and Statistics, 181(1), 277−291, 2016.
[13] Morais E.C., Vigh L.G. and Krähling J., “Preliminary estimation of the probable magnitude of
Komárom 1763 earthquake using fragility functions”, Proceedings of the 16th World Conference
on Earthquake Engineering, Santiago, Chile, (9-13/01), 2017.
[14] Istvánfi G., “Surveys of settlements completed during the summer stages for students of
architecture between 1980 and 1994” (in Hungarian), Institute of Theory and History of
Architecture, BME, Budapest, Hungary, 1995.
[15] Calvi G.M., Pinho R., Magenes G., Bommer J.J., Restrepo-Vélez L.F. and Crowley H.,
“Development of seismic vulnerability assessment methodologies over the past 30 years”, ISET
Journal of Earthquake Technology, 43(3), 75–104, 2006.
[16] Acta Terrae motus anni 1763, Magyar Nemzeti Levéltar (Hungarian National Archive), OL N-98.
Ladula CCC Fascisculus A. No.10.
[17] Genthon I., Hungarian Monuments (in Hungarian), Akadémiai Kiado, 1959.
[18] Rados J., Tata (in Hungarian), Műszaki Könyvkiadó, Budapest, 1964.
[19] Haris A. and Somorjay S., List of Hungarian Monuments: Komárom-Esztergom County (in
Hungarian), Kulturális Örökségvédelmi Hivatal, Budapest, 2006.
[20] Lourenço P.B., Oliveira D.V., Leite J.C., Ingham J.M., Modena C., and da Porto F., “Simplified
indexes for the seismic assessment of masonry buildings: International database and validation”,
Engineering Failure Analysis, 34(1), 585–605, 2013.
[21] Lourenço P.B. and Roque J.A., “Simplified indexes for the seismic vulnerability of ancient masonry
buildings”, Construction and Building Materials, 20(1), 200–208, 2006.
[22] Akkar S. and Bommer J.J., “Empirical equations for the prediction of PGA, PGV, and spectral
accelerations in Europe, the Mediterranean Region, and the Middle East”, Seismological Research
Letters, 81(2), 195-206, 2010.
[23] Vicente F.E. and Torrealva D.E., “Mechanical properties of adobe masonry of historical buildings
in Peru”, Proceedings of the 9th International Conference on Structural Analysis of Historical
Constructions, Mexico City, Mexico, (14-17/09), 2014.

12