1. Introduction
Very often, when assessing the possibility of increasing the live loads in a building, its expansion or even periodic inspection of its technical condition, the mechanical parameters of the structure should be determined. The most appropriate way is to measure them directly during destructive tests (DT) [
1]. Due to the characteristic feature of historical masonry—which is a large heterogeneity of the structure—the number of specimens to be tested and their size must be large. However, the cutting out of masonry prisms is rarely encountered due to the fact that it damages the structure considerably, is expensive and often difficult to implement due to the low fragmentation resistance of the specimen taken. A modern alternative is the flat-jacks test, which is conducted in situ. During the test, the actual stress–strain relationship of the wall is recorded. Despite its versatility and numerous advantages, the use of this method is strongly limited. The necessity to have specialized equipment and personnel with extensive experience in interpreting the obtained results are the most important limitations, also confirmed by the author’s own experience [
2]. A cylindrical specimen of 100–150 mm diameter obtained as boreholes are an increasingly popular alternative [
3]. However, due to technical problems with keeping the specimens undisturbed during the extraction from the wall and the unfavorable influence of water cooling the blade during the drilling, this method requires the researcher to have prior practice.
An easier, and thus often used, method of estimating the compressive strength of a brick wall (as a leading mechanical parameter) is to determine the compressive strength of mortar and bricks separately. The desired compressive strength of the wall is determined on the basis of correlations obtained empirically and published in the literature. This method is especially sanctioned in modern design codes such as EN 1996-1-1 [
4]. The strength of bricks is determined in a very simple way directly on whole bricks taken from the structure or on their halves. In order to reduce the number of bricks to be taken for the study, small cylindrical or cube samples are also used; however, taking into account the anisotropy of the bricks and the influence of scale effect on the results is needed [
3,
5]. It is incomparably more difficult to estimate the compressive strength of the mortar in the existing structure, which results from two basic factors: small joint thicknesses (usually below 20 mm) and degradation of the surface layer of mortar in the joint. The dimensions of the mortar specimen for strength tests required by EN 1015-11 [
6] should be 4 × 4 × 8 cm
3, which excludes the application of the research methodology of this standard in the case of existing buildings. Sometimes few mortar samples of similar dimensions to those required can be extracted from oversized vertical joints of the masonry, where the mortar is, unfortunately, characterized by noticeably lower compaction. Alternatively, the samples of dimensions close to 4 × 4 × 4 cm
3 can be obtained by gluing together three samples (of dimensions 4 × 4 cm
2 in the plane) cut from the bed joint (average 13 mm thick) [
7]. However, the results obtained by this method are understated [
8]. Currently, no general European standard contains a consistent methodology for determining the compressive strength of the mortar based on in situ tests, which is a very undesirable state.
The aim of this article is to evaluate the possibility of using three modern non-standard methods (PT, DPT and TPT) to estimate the compressive strength of the weak mortar in the bed joints of a brick wall. Additionally, an attempt was made to determine the possible limitations of each method by comparing the compressive strength established in all three tests with the reference compressive strength established as described in EN 1015-11 [
6]. Moreover, performing tests several times on specimens of different ages (especially important are those results performed on specimens older than 4 weeks, which is the common total length of the research period in the majority of the articles) allowed for noticing the change in the relationship between the results.
The vast majority of non-destructive and minor-destructive tests are affected by an error due to the fact that the mechanical parameters of the material are not determined directly but by correlation. Extensive studies of this phenomenon were made in [
9,
10]. For example, for the PT test, it is the correlation between the hardness of the mortar and its compressive strength. The biggest advantage of minor-destructive methods is the ability to perform a large number of in situ tests, which is crucial in statistical evaluation. Especially with modern mortars based on recycled powder [
11] or increased porosity [
12]. In addition, the equipment needed to perform these tests is significantly cheaper compared to the machines used for destructive tests. Moreover, most of the MDT allows for immediate interpretation of results.
3. Research Methodology
In order to estimate the range of applicability for the DPT, PT and TPT methods, 3 brick-masonry prisms were built, with dimensions of 51 × 12 × 62 cm
3 each (
Figure 4a). There were 12 × 25 × 6.5 cm
3 bricks used for the prisms. All the bed joints had a fixed constant thickness because of the use of 12 mm spacer strips, which is of particular significance in the case of TPT testing.
Each of the three MDT tests was carried out on three types of mortar (two of them were cement–lime and one was lime) in order to achieve a greater comparative database of all the results (
Figure 4b).
In order to replicate the real conditions in which historical buildings were erected, it was assumed that the components of the mortar would be dispensed by volume—so-called “prescribed mortars”. The ratio values were assumed according to PN-B-10104 [
40] code for F, G and H mortar (
Table 1). In every prism, one bed joint was specifically prepared to allow for the extraction of specimens for the DPT test (
Figure 4c). Applying a separating layer in the form of thin filter paper helped weaken mortar adhesion to the brick. In addition, brick surfaces in this joint were smoothed by grinding. For each of the three types of mortar, three samples 4 × 4 × 16 cm
3 were prepared to determine mechanical parameters according to EN 1015-11 standard [
6]. After 7 days of maturing in high air humidity, the prepared specimens were placed in the same room as the masonry prisms in order to ensure comparable climatic conditions.
The MDT tests were thoroughly comparable due to the fact that each of the masonry prisms was layered with bricks with the use of all three mortars, F, G and H. Additionally, this allowed for increasing the size of the results database. The detailed structure of masonry prisms and the location of the respective MDT tests is presented in
Figure 5. During PT and TPT tests, each of the examined prisms was subject to compression of 0.15 MPa in order to immobilize it.
The PT test was carried out on the outer mortar surface (
Figure 6b) and as the mortar downhole measurements (
Figure 6f). Since the prisms for testing were prepared in the laboratory and not sampled from the existing structure, it was not necessary to remove the outer mortar surface, which may be corroded. At the first PT test stage, there were 20 strokes performed, and the needle probe immersion was recorded after the 5th, 10th, 15th and 20th strokes. As part of the second stage, there was a drill hole made at the same point, with a diameter of 12 mm and a depth of 4 cm in order to conduct the PT test on the mortar located inside the joint (
Figure 6d). The drilling dust was removed from the drill hole by means of compressed air (
Figure 6e). The reading of the needle probe immersion was re-performed after 20 strokes. By using the immersion depth of the needle probe after the 10th stroke, the mortar compressive strength was determined based on the correlation curve, DRC [
29], which was referred to in the author’s previous paper ([
22], Figure 6). On the basis of increments between 5, 10, 15 and 20 strokes the mortar homogeneity was assessed. The methodology was repeated in at least three points, located at a minimum of 10 cm from each other, for each of the tested joints.
The TPT test was performed with the tension wrench, BAHCO (
Figure 3a), with a measurement range up to 35 Nm and a reading accuracy of max. 0.5 Nm. In the first stage of the test, a pilot hole was made with a diameter of 7 mm and a depth of 5 cm. The drill cuttings obtained were checked in terms of the brick particle content, which would disqualify the drill hole from further tests due to the contact of the penetrating nail with the bricks, which are stronger than the mortar. The drill hole diameter was checked, and it was ensured whether, within the entire depth, there were any hollows or large aggregate grains that would distort the results obtained. The drilling dust was removed from the drill hole by means of compressed air.
The DPT test was performed after the completed PT and TPT tests. In order to extract the mortar specimens from three joints for the DPT tests, the subsequent layers of bricks were loosened. The joint-forming mortar was cut into specimens already at the stage of its joining with a brick in order to obtain the specimens with the intact structure as far as possible. Mortar specimen extraction was carried out with the use of a broad chisel. In order to ensure the flat surface of the specimen, it was leveled on both sides with a layer of fast-binding plaster. The leveling layer was obtained by means of steel punches with a diameter of 30 mm, at the same time ensuring the possible correction of the specimen position during the DPT test. For the regular DPT test, steel punches with a diameter of 20 mm were used (
Figure 6q). The tests were performed on the testing machine, ZwickRoell AG (0.5 accuracy class of testing machine with an uncertainty of measurement 0.12%). The load increment in time was selected individually for each of the three mortar types for the specimen destruction between the 2nd and 3rd minute of the test.
The tests of the standard specimens, 4 × 4 × 16 cm
3, were conducted in accordance with the methodology described in EN 1015-11 [
6]. In the first stage, flexural strength was tested (
Figure 6t). The obtained specimen halves, with the dimensions of 4 × 4 × 8 cm
3, were subject to compressive strength tests (
Figure 6u). The tests were carried out with the use of the same testing machine as in the DPT tests.
Repeating the minimally invasive test a few times on the same material is to minimize the possible unfavorable impact of local effects, which may be revealed during the test. Additionally, it increases the number of test results, which enables statistical analysis.
5. Discussion
The ratio of mortar reference strengths
fm.flex/
fm depends both on the binder amount and type and the age of the specimens tested. In the tests carried out on 4 weeks old specimens, the lowest ratio, equalling 0.21, was obtained for the strongest mortar, i.e., F, and the highest ratio, equalling 0.36, was observed for the weakest mortar, i.e., H. Similar values are acknowledged in the literature [
40,
42].
The specimens’ age also had a consistent impact in the case of all the mortar types. The strength ratio achieved the highest value for the tests performed in the 12th week (0.34 for F, 0.61 for G, 0.67 for H), and in the tests in the 90th week, there was a slight decrease (maximum 8% for mortar F containing the most cement). A few factors may have an impact here. The most important one is the difference in the strength increase dynamics
fm.flex in relation to
fm, and the second one is the development of micro cracks as a result of shrinkage. The shrinkage impact on the reduction in flexural strength grows along with the increase in the binder/aggregate ratio [
42]. The slight decreases in the mortar strength over time were also recorded in publications [
41,
44], where this decrease was Δ
fm.flex = −18% and Δ
fm = −21%, respectively.
The comparison of the author’s test results with the results of other researchers (
Figure 15) demonstrates that the relationships of mechanical parameters depend greatly on the mortar composition and curing regime [
42]; numerous correlations were published in [
46]. The mortars tested in this paper are similar to mortars used typically in the existing brick structures—the obtained red curve (in
Figure 15) falls within the dependence devised by Marastoni in [
36], and the black curve describes all the included results. In almost all cases, flexural strength increased more rapidly than compressive strength; the same results were noticed in [
46].
The obtained curves of the mortar compressive strength increase in time, depending significantly on the testing method (see
Figure 16). The highest values were always obtained from the DPT (dotted line) tests, and the significantly lower ones were from the TPT (dash–dot line) tests (except for mortar H, for which no result was obtained). Moreover, the results of the PT tests (dashed line) and the tests performed on reference beam samples (solid line) provided the lowest strengths (
fm for mortars H and F, and
fm.PT for mortar G). Significantly higher values of
fm.DPT result from a few factors, among which the most important are the following:
- (a)
Specimen confinement (the effect of friction at the specimen—punch contact zone; a specimen area larger than the load area; the presence of a capping) [
16,
20,
47,
49,
50].
- (b)
Specimen small dimensions and slenderness (the ratio of specimen thickness to punch diameter) [
8,
16,
17,
22,
26,
51].
- (c)
The beneficial mortar curing regime inside the wall [
20,
52].
The impossibility of the penetrating nail anchoring in mortar H in the TPT test suggests excluding this method from use in the case of very weak mortars. This thesis is confirmed partially by the results obtained by this method’s author [
36], where only one mortar type was tested with a strength of <1 MPa. The differences recorded between the readings were 30%. At the same time, the difference between the mortar strength, determined based on the correlation with the non-destructive test result, ranged from 31% to 80%.
In order to determine the impact of the mortar curing regime, it is necessary to analyze its strength increase dynamics for each of the tests carried out.
Figure 17 presents the above as the comparison of the values between the 4th and the 12th week and the 4th and the 90th week. The highest relative increases (the increase from 0.55 MPa to 2.41 MPa, the ratio of 438%) are characteristic of the PT tests performed on mortar H. This effect probably arises from the highly advantageous coincidence of three factors. The first one, triggered by the presence of bricks, is the reduction in excessive water in the mortar mixture during hardening in the joint. This phenomenon, described in the literature as the “absorption effect”, has a significant impact on the increase in the mortar strength in the actual joints of the structure [
28,
53,
54]. Furthermore, it contributes to the considerable reduction in the total open porosity and the reduction in the average pore radius [
20,
52]. The other factor is fast carbonation which starts in the mortar’s outer surface of the joint, which is where the PT tests [
42] are performed. The third factor is the natural tendency of lime mortars for slow hardening [
46,
48]. To compare, the ratio of reference strength
fm in this period for mortar H was merely 155% (the increase from 0.79 MPa to 1.22 MPa). Regarding mortar F with the prevailing cement share, the highest strength increase was observed for the DPT tests.
Figure 16 and
Figure 17 and the decreases in the total penetration depth (provided in
Section 4.3) in the PT test confirm a strong relationship between mortar strength and its resistance during needle probe penetration. The identical dependence was also confirmed in [
25], where after 6 weeks of curing, the penetration depth was reduced by 15% (cement–lime mortar similar to G) to 21% (lime mortar similar to H). In a similar period (between the 4th and the 12th week), in the tests presented in this paper, the reduction was, respectively, 16% for mortar G and 32% for mortar H. A considerably weaker relationship was obtained in [
28] between mortar compressive strength and the number of strokes.
The minor-destructive test on the mortar in the existing structure is frequently performed in order to use the results obtained for determining the masonry compressive strength [
55,
56,
57]. In such a case, it is often necessary to convert the estimated mortar strength obtained in the MDT test to the mortar compressive strength determined on half-beam specimens 4 × 4 × 8 cm
3 [
58]. The values of the conversion coefficients obtained in the tests for the respective MDTs in the time function are presented in
Figure 18. Nevertheless, their application scope must be limited only to mortars with a similar composition and a testing methodology imitating the one adapted herein. Regarding the TPT and PT tests (with the exception of mortar H), the conversion coefficient value is nearly constant over time. As far as all the DPT tests are concerned (and PT for mortar H), the coefficient values demonstrate a significant increase over time. This is a consequence of the afore-described phenomena.
However, in most cases, the composition of the tested mortar is not known, which necessitates using a function with considerably lower accuracy yet a wider application scope. The dependences obtained in the tests
fm.DPT(
fm),
fm.PT(
fm) and
fm.TPT(
fm) are presented in
Figure 19. Defining a single correlation curve for all three MDT tests (a black line in
Figure 19) is irrational in practice due to the excessively large dispersion of results (R
2 = 0.38).
Regarding the multiplicity of factors affecting the DPT test results (described in detail in the introduction to this section) and, in consequence, considerable variability of ratio
fm.DPT/fm, it would be most advantageous to devise a few curves representing the types of mortars that are used most often. Since papers [
16,
18,
19,
20] provide the detailed results of strength
fm.DPT and
fm, as well as geometrical data for each tested specimen, the dependences between the DPT specimen slenderness and the correlation coefficient, could be elaborated by the author. Said dependences are presented in
Figure 20, with consideration of a function, also elaborated by the author in [
22], dedicated to the historical buildings of the City of Cracow (a black curve).
A continuous line marks the weakest mortar in a given series, a dotted line marks the strongest mortars, and a dashed line marks the intermediate-strength mortars. The strong curvilinear nature of the functions obtained confirms that the correlation coefficient value (fm.DPT/fm) increases as the specimen slenderness decreases (due to the increase in the specimen of the transverse compressive stresses in the radial direction). Another significant factor, which has not been emphasized sufficiently in the literature so far, is mortar strength. The weaker mortars were tested, and the higher correlation coefficient values were obtained. This is also confirmed by the results presented in the diagram in the form of dots recorded for the mortars analyzed herein. The following values of correlation coefficients were obtained for them: 2.53 for F, 3.03 for G and 3.84 for H, despite the comparable slenderness of the DPT specimens.
The correlations between the respective MDTs and DPT tests may be presented successfully with linear functions (
Figure 21). This feature was not dependent on the mortar type.
By analyzing statistical parameters of all results presented in this paper and obtained by other authors [
3,
5,
19,
20,
22,
28,
35,
41,
44,
45,
46,
50,
51,
58,
60,
61], a significant difference between minor destructive tests (DPT, PT, TPT) and the destructive test is noticeable. For dispersion assessment, variation coefficients were compared in
Figure 22. Only reference compressive strength results
fm were characterized by a mean value of CV lower than 10% (precisely 6.2%). All the results of MDTs had this parameter 3–4 times higher. On the other hand, mortar reference strength was tested only on laboratory-prepared specimens with naturally high homogeneity. The value of CV seems to also be affected by mortar type. For all tests performed in this paper, the CV for the strongest mortar (F) was lower than for the weakest mortar (H). The mean value of CV changed from 3% to 12% for
fm, from 11% to 15% for
fm.DPT, from 12% to 26% for
fm.PT, and from 28% to 58% for
fm.TPT. This agrees with the conclusion that more homogenous mortars tend to be stronger [
58]. Those two features suggest that in the case of low-strength mortar diagnostic with PT or TPT, a larger number of tests is needed for the same accuracy of estimation. According to [
62], for practical applications, this dependency seems to become negligible for a number of NDTs measurements equal to at least 20.