The 12th International Conference of
International Association for Computer Methods and Advances in Geomechanics (IACMAG)
1-6 October, 2008
3D Stability Computation of the Praid Salt Mines Complex Using
DKR Control Method
Gy. Deák, Şt. E. Deák, S.O. Mihai
R&D National Institute for Metals and Radioactive Resources, Bucharest, Romania
Keywords: dry salt mining, stability computation, DKRControl method
ABSTRACT: The stability assessment of the complex salt cavities, placed inside the salt rock mass, represents a
real scientific and technical challenge for any researcher focused on the geomechanics field. The purpose of this
paper is 3D stability modelling for the Praid salt mines, Romania, taking into account the salt cavities mining
dynamics and also the intrinsic parameters’ effect over the salt rock mass behaviour (e.g. insoluble and water
soluble residuum contents). FLAC 3D code (Itasca, 2007) was used for numerical computation, in connection with
Kubrix and 3D Shop software, for the geometry and meshing preparation. The proper computation was performed
using DKRControl method principles. The authors developed the preliminary computation, taking into
consideration the exploitation dynamics of the salt mines, the properties of the rock salt determined in the
laboratory, and calibrating the model in accordance with one major effect of the instability phenomenon, well
known and localized in space, such as the roof collapse of the Jozsef Salt mine. The results of this numerical
model exploitation could represent a good start for a detailed analysis of the stability assessment of the Praid salt
cavities complex, at global and especially at local level.
Salt has, perhaps, played the pre-eminent part in the development of European culture and industry. From
earliest times, man and his animals have needed salt in their diet and civilisations have been built and wars
fought over the availability and supply of salt. Salt mines several hundreds years old exist, and even a millennialong
"saline" culture within the border of the Romanian area, in the same time with nowadays active mining sites.
Long before modern mining methods were developed, huge salt underground cavities were created, of enormous
and impressive sizes (more than 90 m in height and 20 m in width). These mines are under threat of collapsing as
long-term geotechnical and hydro-geological conditions are changing and indeed many have already collapsed.
Praid salt mine is one of these sites. The aim of the Praid saline complex global stability computation is to
diagnose problematic areas for salt mine museums as cultural heritages and to determine the long term influence
over such of the mining activity in progress at the Mina Noua and Telegdy salt mines, in order to provide the
means and solutions for efficient conservation and sustainable exploitation for both, man-made and natural salt
2 General presentation of the Praid salt mine
The Hill of the Salt (the highest point – 576 m), covers the biggest salt body from the country, with 2.7 ÷ 3.0 km
development in depth. In horizontal section, the salt diapir has an elliptic form, with diameters of 1.2 ÷ 14.0 km
and in vertical section it has a huge mushroom shape. This irregular form is a characteristic feature of the salt
bodies from Transylvania, Romania (Horváth, 2005). The salt from Praid is very old (Badenian inferior age ÷ the
middle Miocene age), sometimes 20 - 22 million years ago. This salt settlement is one of the biggest from
Europe. Its reserve could provide the salt need for Europe for hundreds of years in future, for which reason it is
nicknamed Europe’s salt cellar.
The beginning of the salt mining history lasts ever since the Roman period. In written documents there are
remembered four surface excavations ("amphitheatre" type) and some bricks with the inscription: LVM (Legio V.
Macedonica). The systematic underground exploitation began in 1762, when the József mine was opened in the
South-West side of the Salt Hill. This mine had a bell-shape, Figure 1, and it was one of the first underground
mines in Praid.
Figure 1. Bell-shape room, Transylvanian salt mine
(after Minero-logia principatus Transsilvaniae, Fridvalski, 1787).
From the bell-shaped hall of József mine, two side rooms were opened; these were the so called Károly and
Ferdinánd mines, which also had pointed shape. Their depth was of 66m, together with the shaft. In 1864 near
the József mine the Parallel mine was opened, which was one of the biggest artificial salt cavities in Romania, of
100 m height. This mine was the first trapezoid shaped salt room, and in its wall the annual deep level and also
the names of the "mine giants" (the leader engineers) were engraved. Between 1947 ÷ 1949 the Dózsa mine was
opened and in 1978 at 40 m under the old mines, new mining levels were opened. In those levels the extraction
rooms were 20 m width, 12 m high and 100 m long. In 1991 the opening of a new mining sector began, the
Telegdy mine. The continuous mining activity began here in 1994. In this sector the Canadian method is applied,
with small room dimensions of 8 m high, 16 m width and quadric "pillars" of 14 m × 14 m. Nowadays the deepest
exploitation is carried out at 320 m depth, measured from the surface.
3 DKRControl method principles
DKRControl method is a complex engineering tool to solve geo-mining situations with determined level of
confidence (Deák and Deák, 2005). As a matter of fact, DKRControl method makes the connection between three
important items: geo-mining problem (informational volume), numerical computation (FLAC code) and
engineering solutions (Deák et al., 2004). This is the first application of the DKRControl method for 3D stability
Step 1: Analysis of the actual informational volume
All six informational volumes were used to define the geo-mining problem: geology, exploitation dynamics, rocks
properties, in situ measurements, indirectly data and qualitative data.
Step 2: Establishing a strategy to improve the knowledge level
This step ensures a high level of confidence concerning the solving of a particular geo-mining problem, from two
points of view: global stability approach and local stability approach (due to the local low intensity phenomena). It
consists in improving or at least preserving the informational volume.
Step 3: Numerical model Parajdi-Sóbányák-3D
Numerical model geometry and meshing were performed in this step. It makes a delimitation of those areas of
rock salt mass, characterized by the same intrinsically properties, to ensure a constant geo-mechanical and
rheological salt behaviour. The conventional variation of basic properties is maximum 25%, in the same interval of
sterile content level. The recommended constitutive equation of the model is for elasto-viscous behavior, as it
( ) Σ=
= ⋅ +
k ,t k k
a 1 α
where ε is the vertical strain, %; a is the elasticity parameter; δ , α are the rheological parameters; m is the
content of insoluble and soluble material; k = x or y or z are the main directions; t is the time, hours; i, n are serial
numbers. All these parameters are necessary for triaxial tests.
Step 4: Achievement of a control system over the studied phenomenon
The control system should be based on (Deák, 2003; Hagedorn and Konietzky, 2007):
- in situ measurements for one or more effects of the geomechanical phenomenon,
- the correlation between dynamic of exploitation / excavation and instability phenomena (known and
- the correlation between the dynamics of the exploitation / excavation and the chosen control system,
- in space and time correlation between the chosen control system and geometry of the numerical model.
Step 5: Calibration of the numerical model
Step 6: Validation of the numerical model
Step 7: Numerical simulation of geo-mining phenomena
Step 8: Assessment by computation of the studied geo-mining phenomena
4 3D numerical model for the Praid salt cavities complex
Grounded on the existent informational volume, the authors have recreated the working geometrical model that is
the most consistent with the existent topographical situation of the studied area, as of the 15th of September
2007. Figure 2(a) shows the initial working geometry. Figure 2(b) shows the working geometry after the levelling
of the surface soil, such that the loading resulting from the rocks column to be higher than or equal to the real
one, however under the condition that the variations should not exceed 25%.
Figure 2. Numerical model geometry.
The main idea consisted in the fact that the shaping of the topographical surface was made upon recent
measurements, also including the subsistence phenomenon generated by the salt cavities and saline erosions
existent on the salt peaks. In the same time, inside the massif, the geometry of the salt cavities does not take into
consideration in this computation step the opening and preparing mining works, such as shafts, adits and
ventilation pits. Consequent to the decisional qualitative analysis of the informational volume, according to step 1,
the authors have reached the conclusion that the trust level reaches in this step an acceptable limit. Without an
improvement program, only a preliminary assessment can be performed regarding the saline stability and the
mutual influence of salt cavities. For step 2, in the future there must be performed laboratory investigations such
as triaxial tests and in situ geophysical tests, namely the enrichment of the informational volume regarding
topographical measurements and DHG experimental offset measurements, Figure 3.
Figure 3. DHG experimental offset.
In Step 3, compared to 2D modelling, where by means of the dek parameter (dek parameter defined in the space
the boundary areas from the elastic to plastic and rheologic behaviour of the rock mass, according to the material
properties, determined in the laboratory, under triaxial loading) there could be controlled the shift of an area from
the elastic towards the plastic and rheological status, the difference in the 3D modelling, for now, is that the dek
parameter only controls the shift from the elastic status towards the plastic one. The actual numeric model was
created, in FLAC 3D code, as it follows:
Set geometry 1e-7
CONFIG GPEX 9 ZEX 12
Impgrid Parajdi-sobanyak-mohr (performed with Kubrix and 3D Shop software)
pro bulk 1.6913e11 ga 4.6362e10 she 1.5201e10 ga 6.2788e9
fix x y z range z 179 180.2
fix x y range x 339,400.5 y 500,615 z 182,565
fix x y range x 100,165 y -111,-55 z 182,565
fix x y range x -111,-30 y 100,330 z 182,565
fix x y range x 540,614 y 300,350 z 182,565
fix x y z range x -110.8,-109.5 y 329.5,330.9 z 481.8,482.5
fix x y z range x 338.8,339.5 y 609.8,610.5 z 563.8,564.5
fix x y z range x 613.8,614.5 y 169.9,170.5 z 509.8,510.5
fix x y z range x 164.7,165.5 y -110.9,-111.7 z 486.8,487.5
fix x y range x -30, 30 y 350,420 z 182,565
fix x y range x 0,50 y 0,140 z 182,565
ini dens 2100
set g 0 0 -10
Initialize gravity stresses
pnt = zone_head
loop while pnt # null
z_depth = z_zcen(pnt)
SS_mod = 10*dens *(-SUP + z_depth)
z_SXX(pnt) = SS_mod *K_RATIO
z_SYY(pnt) = SS_mod *K_RATIO
z_SZZ(pnt) = SS_mod
pnt = z_next(pnt)
if … than
..mohr & prop
ini xdisp=0 ydisp=0 zdisp=0
Steps 4, 5 and 6 were made grounded on the exploiting dynamics, knowing the main instability phenomena
registered in the salt mines and localized in space. Figure 4 shows the exploitation dynamics of the salt mines,
reconstituted based on the existent documentation, starting with year 1762 and until the present moment,
meaning a time interval of 245 years.
Figure 4. Exploiting dynamics of the salt mines.
5 Numerical model computation
The numerical model comprises the salt rock mass and another 12 components , detailed in Table 1.
Table 1. Numerical model content
Issue Group Number of the
Number of the
Order of the cavity
0 Salt rock mass 83109 92984 not applicable
1 Jozsef Salt Mine 65 174 the first one
2 Parhuzamos Salt Mine 118 288 the second one
3 Dozsa Salt Mine 240 532 the third one
4 426-So Kamra 40 120 the fourth one
5 Level-402/Salt Sanatorium 340 830 the fifth one
6 Level +375 m 261 678 the sixth one
7 Level +339 m 649 1614 the seventh one
8 Level +286 m, Mina Noua Salt Mine 429 1114 the eighth one
9 Level +266 m, Mina Noua Salt Mine 381 1028 the ninth one
10 Level +246 m, Mina Noua Salt Mine 605 1460 the tenth one
11 DHG Offset*, Mina Noua Salt Mine 2 12 the tenth one
12 Level +226 m, Mina Noua Salt Mine 452 1144 the eleventh one
* DHG offset is included in the model just because will provide, in time, valuable in situ measurements, for calibration and
The numerical model exploitation was performed in 12 cycles, indicated by the sequence:
model null ra group name_elas (ex Jozsef_banya)
save name_plas.sav (ex Jozsef_banya.sav)
The results of the numerical computation, based on the geomechanical properties, determined in laboratory,
indicate the salt cavities complex is stable, at global level, but the real situation is different. Taking into
consideration the actual effects of the geomechanical phenomena, the authors decided that is necessary to
determine the modality to readjust the geomechanical properties, determined in laboratory, in accordance with the
rock mass scale behaviour. The solution consisted in the numerical model calibration according to the roof
collapse of Joszef salt mine.
The numerical computation was consecutive calibrated for five times. Figures 5(a) ÷ 5(e) illustrate the failure zone
evolution, respectively the roof collapse of the Jozsef salt mine, Figure 6 (qualitative information).
Using the qualitative informational volume of the Jozsef salt mine, the numerical model of the Praid salt cavities
complex, Figure 7, was preliminary calibrated. After the five calibrations and computation stages, the resulted
global stability of the Praid salt cavities complex is still good, except some local phenomena, like spalling and
exfoliation rock, Figure 8. The trend of these phenomena indicates an increasing of their amplitude in time. The
most exposed zone is the area located below the central pillar of the old salt mines and the salt cavities located
deeper than 300 m.
Except Jozsef salt mine, affected by the roof collapse, the other old salt mines have a good stability, according to
the actual level of confidence, possible basis for further actions to recover them as salt museums, sanatoria and
Figure 5. Numerical computation of the roof collapse, Jozsef salt mine.
Figure 5(e). Computed salt roof, Josef salt mine. Figure 6. Roof collapse of the Josef salt mine.
Figure 7. Praid salt mine complex before the first computation stage
Figure 8. Results of the preliminary stages of the numerical computation,
Praid salt mine complex.
7 Final considerations
The numerical model of the Praid salt cavities complex, named Parajdi-Sóbányák-3D, comprises 83109 zones of
the 12 issues. In order to devlop the geometry of this complex model, proper for numerical computation using
FLAC 3D code, the Kubrix and 3D Shop software facilitation were necessary. The authors mention these facilities
are useful, but not sufficient in order to perform an adequate meshing for the global stability computation.
Because of the complex structure of the model, it was almost impossible to quantify the local instability effects.
The authors developed the preliminary computation, taking into consideration the exploitation dynamics of the salt
mines, and calibrating the model in accordance with one major effect of the instability phenomenon, well known
and localized in space, such as the roof collapse of the Jozsef salt mine. This qualitative calibration is not enough
to get results with high level of confidence (e.g. more than 80%), but it is useful for efficiency improving of a such
complex numerical model.
Preliminary conclusions, provided by the numerical computation, indicated for the old salt mines, Parallel Salt
Mine (138 years) and Dozsa salt mine (59 years) have a very good stability, in spite of their age. The other salt
mines, except Jozsef salt mine, have a global good stability, but indicate local instability problems (e.g. spalling
and exfoliation rock of the corners, roof or pillars) with high intensity, which can be real sources of the active salt
mines collapse, according to the domino principle. In particular, the sites placed below the central pillar of the old
salt mines, as well as the Jozsef salt mine are exposed to this threat.
The authors recommend the numerical model meshing improving, the rheologic computation, and the model
calibration and validation using in situ measurement results (e.g. surveying, displacements in DHG offset area,
etc. – used techniques of DKRControl method, Deák et al., 2004).
The authors wish to acknowledge that this paper is based on data provided by the specialists of the Praid Salt
Mine. The authors also want to thank Ms. Michele Nelson for her gentles in providing the Kubrix and 3D Shop trial
licenses for this complex modelling.
Deák Gy. 2003. Monitoring and control system for stability conditions improving of dry salt rock mining, Doctoral thesis,
University of Petrosani, Romania
Deák Gy., Deák Şt.E., Georgescu M., Predescu C., Scurtu M., Cioata C. 2004, Mining reinforcement viability determination in
situ in strong tectonized rock mass using DKRControl method, Proc. 1st Int. UDEC/3DEC Symp., Bochum (Germany), 219-
Deák Gy., Deák Şt.E. 2005. Collapse risk evaluation of the salt dissolution or dry mining using DKRControl method. Proc. 1st
Int. Seminar ECOMINING-Europe in 21st Century, Sovata, (Romania), 121-127.
Fridvalski J. 1787. Minero-logia magni principatus Transilvania seu Metalla, Semi-Metalla, Sulphura, Salia, Lapides & Aqua
conscripta, Claudiopoli, Typis Academicis Societis Jesu.
Hagedorn H., Konietzky H. 2007. Numerical Modeling of rock bursts in the multifunction station Faido of Gothard Base Tunnel.
Proc. 2nd Int. Seminar ECOMINING-Europe in 21st Century, Sovata, (Romania), 127-138.
Horvát I. 2005. Mining of the Praid salt mine. History in brief (in Romanian), Lyra, Tg. Mures, Romania.
Itasca Consulting Group, Inc. 2006. FLAC 3D - Fast Lagrangian Analysis of Continua in 3 Dimensions Ver. 3.1, Command
Reference, Minneapolis, Minnesota (USA).
The 12th International Conference of