In this present study, a comprehensive dataset comprising 160 experimental tests was gathered. The tests involved single-storey, single-span reinforced concrete frames with masonry infill, which were subjected to lateral loading. The primary objective is to conduct a statistical analysis focusing on the strength and stiffness characteristics of the masonry infill. For a visual representation of the collected experimental tests, please refer to Fig. 2, which provides a concise summary of the collected data.
The references for the collected tests, as well as the type of loading applied and the number of tests performed, are provided in Table 1
Table 1
References of the different tests collected for the infilled frame structures
nº | Test references | Loading type | Number of tests |
1 | Akhoundi et al. [40] | Quasi-static cyclic | 6 |
2 | Alwashali et al. [41] | Quasi-static cyclic | 5 |
4 | Basha et Kaushik [42] | Quasi-static cyclic | 9 |
5 | Bergami et Nuti [43] | Quasi-static cyclic | 2 |
6 | Butenweg et al. [44] | Quasi-static cyclic | 3 |
7 | Cavaleri & Di Trapani [45] | Quasi-static cyclic | 12 |
8 | Colangelo F. [46] | Pseudo-dynamic | 11 |
9 | Da porto et al. [47] | Quasi-static cyclic | 6 |
10 | Dautaj et al. [48] | Quasi-static cyclic/ Monotonic | 7 |
11 | Gazic & Sigmund [49] | Quasi-static cyclic | 11 |
12 | Huang et al. [50] | Quasi-static cyclic | 4 |
13 | Kakaletsis et al. [51][52] | Quasi-static cyclic | 4 |
14 | Layadi et al. [53] | Monotonic | 2 |
15 | Maidiawati et al. [54] | Quasi-static cyclic | 3 |
16 | Mehrabi & Shing [55] | Quasi-static cyclic | 13 |
17 | Misir et al. [56] | Quasi-static cyclic | 2 |
18 | Misir et al. [57] | Quasi-static cyclic | 5 |
19 | Morandi et al. [58] | Quasi-static cyclic | 5 |
20 | Schwarz et al. [59] | Quasi-static cyclic | 7 |
21 | Sigmund & Penava [60] | Quasi-static cyclic | 9 |
22 | Suzuki et al. [61] | Quasi-static cyclic | 4 |
23 | Tawfik Essa et al. [62] | Quasi-static cyclic | 3 |
24 | Tekeli & Aydin [63] | Quasi-static cyclic | 9 |
25 | Van & Lau [64] | Quasi-static cyclic / Monotonic | 4 |
26 | Verdame et al. [65] | Quasi-static cyclic | 2 |
27 | Zhai et al. [66] | Quasi-static cyclic | 3 |
28 | Zovkic [67] | Quasi-static cyclic | 9 |
| Total number of collected tests | | 160 |
From the collected tests (160 tests), certain tests were removed due to their involvement with out-of-plane loading, confined masonry, or reinforced masonry infills and masonry infills with openings. As a result, only 119 tests were retained from the initial database. A summary of the masonry infill types present in the database can be found in Table 2.
Table 2
Types of masonry used in the database
Masonry unit type | Number of samples |
Hollow Clay brick | 55 |
Solid Clay brick | 20 |
Hollow concrete bloc | 16 |
Concrete bloc | 7 |
AAC | 7 |
Calcareneite brick | 4 |
Fly ashes brick | 9 |
Ceramic hollow | 2 |
Pumice bloc | 1 |
As can be observed, the collected tests include the loading type of monotonic loading. Van and Lau [64] demonstrated that, in terms of strength, there was no significant difference in the Infilled frame structures between cyclic and monotonic loading types. However, a noticeable difference was observed in terms of stiffness, with frames subjected to cyclic loading exhibiting higher stiffness compared to those under monotonic loading. Based on these observations, it was decided to incorporate both types of loading for strength determination, while excluding tests conducted under monotonic loading for stiffness determination. In regards to pseudo-dynamic loading, it is important to note that this present study assumes negligible differences in strength and stiffness between cyclic loading and pseudo-dynamic loading for masonry infilled frames. However, it is important to recognize that this assumption represents a limitation of the study.
During the collection of experimental data, several pieces of information, particularly regarding the mechanical characterization of the masonry infill, were not provided by the authors. These include the modulus of elasticity of the masonry (Ew), masonry shear strength (fwu), and in some cases, masonry compressive strength (fwv).
2.1. Evaluation of the masonry compressive strength
To characterize masonry in compression, two types of tests can be utilized: the prism compression test and the wallette compression test (refer to Fig. 3). Furthermore, to enhance our understanding of masonry behaviour under compression, a separate database was collected, consisting of masonry compressive tests. This database comprises 260 mean values derived from 1133 individual tests. Notably, this updated database surpasses the ones previously gathered by Sarhat & Sherwood [68], which included 248 mean values from 1092 individual tests, as well as the database compiled by Thaickavil & Thomas [69], which comprised 232 mean values.
The database used in our study is notable for its diverse range of test types. Within the collected database, there are 63 average values derived from tests conducted on wallettes, while 197 values originate from tests performed on prisms. This stands in contrast to the databases gathered by Sarhat and Sherwood [68], as well as Thaickavil and Thomas [69], which predominantly focused on prism tests with only a few cases of wallette tests. Indeed, recent studies [70] [71] have indicated that wallette tests consistently yield significantly lower strengths compared to prism tests.
Table 3 summarizes the references of the collected tests, along with their mechanical characteristics and the type of test conducted.
Table 3
References of the collected tests to characterize the masonry in compression
References | Material | fb (MPa) | fm (MPa) | Type of test |
Basha & Kaushik [72] | - Fly ashes | 5.7 | [6.9–21.6] | Prism |
Bennett et al. [73] | - Clay hollow | 35.6 | 16.7 | Wallette |
Bergami [74] | - Clay hollow | [4.97–23.39] | [11.72–23.49] | Wallette |
Bustos-García et al. [75] | - Clay solid | 45 | 5.2 | Wallette |
Caldeira et al. [13] | - Concrete hollow | [16.3–45.6] | [5.6–18.3] | Prism |
Calderon et al. [71] | - Clay hollow | [19.5–22.3] | [7.57–27.98] | Wallette & Prism |
Cavaleri et al. [76] | - Clay hollow - Calcareinite bloc - Concrete hollow | [4.07–37.68] | [3.06–9.89] | Wallette |
Da Porto et al. [77] | - Clay hollow | [23.49–25.15] | [11.51–17.68] | Wallette |
Da Porto et al. [78] | - Clay hollow | 13.5 | 19.9 | Wallette |
Esposito et al. [79] | - Clay hollow - Calcium silicate brick | [16–25] | [6.11–7.24] | Wallette |
Ferreti et al. [80] | - AAC | 3.1 | 7 | Wallette & prism |
Furtado et al. [81] | - Clay hollow - Concrete hollow | [1.5–3.25] | [5.54–13.5] | Wallette |
Gumaste et al. [82] | - Table mounted bricks - Wire cut bricks | [5.7–23] | [0.86–12.21] | Wallette & prism |
Kashik et al. [83] | - Clay solid | [16.1–28.9] | [3.1–20.6] | Prism |
Lumantarna et al. [14] | - Clay solid | [8.5–43.4] | [0.69–23.2] | Prism |
McNary, & Abrams [84] | - Clay solid | [69.8–101.7] | [3.4–52.6] | Prism |
Monteagudo et al. [10] | - Clay solid | 59.8 | 22.9 | Wallette & prism |
Morandi et al. [85] | - Clay hollow | 8.64 | 7.68 | Wallette |
Padalu & Singh [86] | - Clay solid | 26.16 | 14.75 | Prism |
Penava et al. [87] | - Clay hollow | 14.5 | 5.02 | Wallette |
Radovanović et al. [88] | - Clay hollow - Concrete hollow | [3.26–6.51] | [8.2–23.9] | Wallette |
Sarangapani et al. [89] | - Clay solid | [3.17–10.67] | [4.2–10.57] | Prism |
Silva et al. [90] | - Clay hollow | 14.05 | 10.38 | Wallette |
Singhal & Rai [91] | - Clay solid | [21.9–40] | 8.5 | Prism |
Singh & Munjal [92] | - Clay solid - Concrete solid | [8.24–16.71] | [12.66–20.85] | Prism |
Thaickavil & Thomas [69] | - Cement stabilised pressed brick - Clay solid | [4.56–6.68] | [13.6–35.5] | Prism |
Thamboo & Dhanasekar [70] | - Clay solid - Compressed earth brick | [3.8–15.8] | [3.98–6.46] | Wallette & Prism |
Veríssimo-Anacleto et al. [93] | - Stone brick | 36.49 | 6.73 | Wallette |
Wang et al. [94] | - Clay solid | 8.4 | 32.6 | Wallette & prism |
Wu et al. [95] | - Hollow shale and coal gangue | 5.55 | 5.34 | Wallette |
Zhou et al. [96] | - Concrete hollow | [23.15–36.75] | [5.6–13.73] | Prism |
Zovkić [67] | - Clay hollow - Concrete hollow - AAC | [2.12–13.21] | [4.04–13.89] | Wallette |
2.2. Relationship between elastic modulus and the masonry compressive strength
In literature, various recommendations have been made regarding the Ew/fwv ratio. However, different values have been reported in different sources. For instance, Eurocode 6 [97] suggests a value of 1000, the Canadian code CSA: S304.1 [98] suggests 850, the MSJC 2013 [99] suggests 700, and the FEMA 306 [100] suggests 550. Several authors have also proposed their own values for this ratio. Calderon et al. [71] propose a value of 580, while Wang et al. [94] consider the average of the recommendations from [101] and use a value of 900. Caldeira et al. [13], based on the collected data, state that the range of values found in the literature is too dispersed to draw conclusive general trends. Table 4 provides a summary of the recommended values found in the literature.
Table 4
Relationship between modulus of elasticity and compressive strength of masonry
Reference | Equation | Reference | Equation |
[83][100] | Ew = 550 fwv | [97] | Ew = 1000 fwv |
[102] | Ew = 500 fwv | [98] | Ew = 850 fwv |
[71] | Ew = 580 fwv | [99] | Ew = 700 fwv |
[13] | Ew = 474.5 fwv | [104][105] | Ew = 750 fwv |
[103] | Ew = 1180 fwv 0.83 | [72] | Ew = 600 fwv |
[94] | Ew = 900 fwv | [106] | Ew = 800 fwv |
Based on the collected tests, only 157 tests evaluated masonry stiffness, specifically 47 wallette tests and 110 prism tests. Through statistical analysis, it was found that wallette tests yielded an average stiffness value of Ew = 1080 fwv (Cov = 75.66%). This value is relatively close to the recommendations provided by EN 1996-1-1 [97], which are based on tests conducted on wallettes. On the other hand, the prism tests resulted in an average stiffness value of Ew = 540 fwv (Cov = 49.73%), which aligns closely with the recommendations of FEMA 306 [100]. It's important to note that the coefficients of variation obtained are relatively high, particularly for the wallettes. Therefore, relying only on a simple average would lead to significant inaccuracies (see Fig. 4).
As shown in Fig. 4(a), the range of values for the Ew/fwv ratio is much wider in the wallette tests compared to the prism tests depicted in Fig. 4(b). The wallettes exhibit values ranging from 400 to 2500, whereas the prisms have a narrower range of 350 to 800, despite having a higher number of tests conducted. Conducting a linear regression analysis, we obtained values of 695 (R² = 68%) for wallettes and 426 (R² = 87%) for prisms. In a study by Thamboo and Dhanasekar [70], they observed that wallettes displayed lower values of elastic modulus compared to prisms, while Calderon et al. [71] reported opposite results. However, our study, based on a larger database, allows us to assert that the prism tests indicate lower values compared to the wallette tests.
Gumaste et al. [82] have shown that prism tests exhibited a greater dispersion of strength values compared to wallette tests, indicating that wallette tests were more reliable. However, in this present study, we observe the opposite phenomenon concerning the modulus of elasticity values. Wallette tests demonstrate a substantial dispersion, whereas prism tests exhibit a certain concentration around specific values (refer to Fig. 4). Based on these observations, it appears appropriate to utilize prism tests for assessing masonry stiffness, while wallette tests are more suitable for evaluating its compressive strength.
For this present study, an average value of Ew = 560 fwv is retained, which is obtained by averaging the values from the linear regressions for the two types of tests. This value is consistent with numerous recommendations found in the literature [59] [83] [100] [71] [72], and it is also close to the average stiffness displayed by the prism tests
2.3. Empirical Relationship for Assessing Masonry Compressive Strength
The prediction of compressive strength has been extensively studied, the topic has been the subject of numerous papers. Some authors have employed artificial neural networks (ANN) [107] [96] [108] to predict the compressive strength of masonry walls. Meanwhile, other authors have proposed empirical equations based on linear or non-linear regression. Table 5 presents the most reliable equations found in the existing literature
Table 5
Empirical models for assessing the compressive strength of masonry
Empirical model reference | Empirical model |
Engesser [109] | \({f_{wv}}=\frac{1}{3}\,\,{f_b}+\,\,\frac{2}{3}\,{f_m}\) |
Mann [110] | \({f_{wv}}=0.83\,{f_b}^{{0.66}}\,\,{f_m}^{{0.18}}\) |
Kaushik et al. [83] | \({f_{wv}}=0.63\,f_{b}^{{0.49}}\,\,{f_{mc}}^{{0.32}}\) |
Lumatarna et al. [14] | \({f_{wv}}=0.75\,f_{b}^{{0.75}}\,\,f_{m}^{{0.31}}\) |
Kumavat [111] | \({f_{wv}}=0.69\,f_{b}^{{0.6}}\,\,f_{m}^{{0.35}}\) |
Garzón-Roca et al. [108] | \({f_{wv}}=\,\,0.53\,{f_b}\,+0.93\,\,{f_m} - 10.32\) |
Gumaste et al. [82] | For prism \({f_{wv}}=0.317\,f_{b}^{{0.866}}\,\,f_{m}^{{0.134}}\) For wallette \({f_{wv}}=\,\,1.242\,\,f_{b}^{{0.531}}\,\,f_{m}^{{0.208}}\) |
Christy et al. [112] | \({f_{wv}}=\,\,0.35\,\,f_{b}^{{0.65}}\,\,f_{m}^{{0.25}}\) |
Bennett et al. [73] | \({f_{wv}}=0.3\,\,{f_b}\) |
Basha & Kaushik [72] | \({f_{wv}}=\,\,1.34\,\,f_{b}^{{0.1}}\,\,f_{m}^{{0.33}}\) |
Thamboo & Dhanasekar [70] | For prism \({f_{wv}}=\,\,0.2\,\,f_{b}^{{1.26}}\,\,f_{m}^{{0.15}}\) For wallette \({f_{wv}}=0.25\,\,f_{b}^{{1.09}}\,\,f_{m}^{{0.12}}\) |
Bröcker [113] | \({f_{wv}}=0.68\,\,{f_b}^{{0.5}}\,\,{f_m}^{{0.333}}\) |
Dymiotis & Gutlederer [114] | \({f_{wv}}=0.3266\,\,{f_b}\left( {1 - 0.0027\,\,{f_b}+0.0147\,\,{f_m}} \right)\) |
Calderon et al. [71] | \({f_{wv}}=\,\,\frac{{2.042\,\,{f_b}^{{0.193}}\,\,{f_m}^{{0.549}}}}{{{{\left[ {\frac{{\left( {{\raise0.7ex\hbox{$h$} \!\mathord{\left/ {\vphantom {h t}}\right.\kern-0pt}\!\lower0.7ex\hbox{$t$}}} \right)}}{3}} \right]}^{0.166}}}}\) |
Hendry & Malek [115] | \({f_{wv}}=0.317\,\,{f_b}^{{0.531}}\,\,{f_m}^{{0.208}}\) |
Dayaratnam [116] | \({f_{wv}}=0.275\,\,{f_b}^{{0.5}}\,\,{f_m}^{{0.5}}\) |
Veríssimo-Anacleton et al. [93] | \({f_{wv}}=\,\,\,\left( {1 - {e^{ - \left( {0.059\,\,x\,\,{n^{0.205}}\,x\,7.943\,\,\,{h^{0.3}}} \right)}}} \right)\,\,\,\left( {0.193\,\,{f_b}^{{1.3}}\,\,{f_m}^{{0.28}}} \right)\) |
In the existing literature, there is a lack of comparative studies evaluating the performance of these equations in predicting masonry resistance using prism and wallette tests. Previous authors have primarily focused on prism tests [86] [69]. However, considering the findings presented in [70] and [71], conducting a comparative study appears to be highly relevant.
Figure 5 provides a comparison of the 17 empirical models presented in Table 5. Additionally, a regression analysis was conducted, resulting in the derivation of Eqs. (3) and (4) for the prism and wallette tests, respectively.
$${f_{wv}}=\,\,0.155\,\,f_{b}^{{1.05}}\,\,f_{m}^{{0.236}}$$
3
$${f_{wv}}=0.35\,\,f_{b}^{{0.891}}\,\,f_{m}^{{0.133}}$$
4
Figure 5 and 6 illustrate the comparison of the empirical model's predictions with experimental test results. The diagonal line on each graph represents perfect prediction, while the scattered points represent the values obtained for each model. A point below the diagonal line indicates an underestimation of resistance by the model, whereas a point above indicates an overestimation. To facilitate comparison between the models, statistical parameters were calculated and are presented in Table 6. The table is divided into two sections, one for the models evaluated using prism tests and the other for those evaluated using wallette tests.
Table 6
Evaluation of empirical models on (a) prism tests (b) wallette tests
(a) | Average* | Min* | Max* | SD | Cov (%) | MAD | RMSE | MAPE | MPE |
Equation (3) | 1.00 | 0.287 | 2.542 | 0.449 | 44.9 | 2.273 | 3.674 | 36.2 | 0.305 |
Engesser [109] | 5.15 | 0.655 | 27.08 | 5.310 | 103.0 | 9.037 | 10.93 | 418.3 | 415.47 |
Mann [110] | 1.92 | 0.470 | 5.314 | 1.186 | 61.6 | 3.024 | 3.848 | 105.1 | 92.556 |
Kaushik et al. [83] | 1.38 | 0.273 | 4.814 | 1.107 | 74.7 | 3.741 | 6.049 | 85.2 | 48.122 |
Lumatarna et al. [14] | 2.97 | 0.731 | 9.061 | 1.798 | 60.4 | 6.234 | 7.969 | 199.1 | 197.560 |
Kumavat [111] | 2.21 | 0.452 | 7.232 | 1.542 | 69.8 | 3.517 | 4.186 | 133.2 | 120.921 |
Garzón-Roca et al. [108] | 3.39 | -5.19 | 27.00 | 4.984 | 146.8 | 8.048 | 11.07 | 308.8 | 239.526 |
Gumaste et al. [82] | 1.03 | 0.308 | 2.619 | 0.512 | 49.8 | 2.830 | 4.733 | 42.1 | 2.791 |
Christy et al. [112] | 0.95 | 0.228 | 2.823 | 0.611 | 64.0 | 3.890 | 6.606 | 52.4 | -4.520 |
Bennett et al. [73] | 0.94 | 0.281 | 2.299 | 0.405 | 42.8 | 2.555 | 4.379 | 34.3 | -5.478 |
Basha & Kaushik [72] | 1.49 | 0.106 | 5.446 | 1.407 | 93.9 | 5.174 | 8.335 | 108.2 | 49.845 |
Thamboo & Dhanasekar [70] | 1.06 | 0.310 | 2.761 | 0.441 | 41.7 | 2.187 | 3.517 | 35.1 | 5.629 |
Bröcker [113] | 1.69 | 0.317 | 5.553 | 1.262 | 74.6 | 3.572 | 5.481 | 97.8 | 69.119 |
Dymiotis & Gutlederer [114] | 1.21 | 0.376 | 3.156 | 0.552 | 45.5 | 2.126 | 3.365 | 44.7 | 21.363 |
Calderon et al. [71] | 4.96 | 0.317 | 18.747 | 4.698 | 94.6 | 8.561 | 9.919 | 404.9 | 396.50 |
Hendry & Malek [115] | 0.60 | 0.120 | 1.755 | 0.418 | 69.3 | 5.086 | 8.581 | 51.8 | -39.731 |
Dayaratnam [116] | 1.07 | 0.128 | 4.072 | 0.863 | 80.2 | 4.319 | 7.041 | 69.7 | 7.563 |
Veríssimo-Anacleton et al. [93] | 1.88 | 0.339 | 6.327 | 0.986 | 52.4 | 5.051 | 12.44 | 98.3 | 88.342 |
(b) | Average* | Min* | Max* | SD | Cov (%) | MAD | RMSE | MAPE | MPE |
Equation (4) | 1.01 | 0.485 | 2.350 | 0.384 | 38.1 | 1.267 | 1.894 | 28.8 | 0.889 |
Engesser [109] | 3.13 | 0.842 | 10.823 | 2.090 | 66.6 | 6.244 | 7.652 | 214.1 | 213.52 |
Mann [110] | 1.57 | 0.653 | 3.241 | 0.537 | 34.2 | 1.916 | 2.570 | 59.48 | 57.054 |
Kaushik et al. [83] | 1.11 | 0.369 | 2.372 | 0.445 | 40.0 | 1.355 | 2.135 | 33.58 | 11.324 |
Lumatarna et al. [14] | 2.25 | 0.998 | 4.689 | 0.811 | 35.9 | 4.827 | 6.411 | 125.6 | 125.63 |
Kumavat [111] | 1.63 | 0.609 | 3.407 | 0.605 | 37.1 | 2.113 | 2.864 | 65.78 | 63.04 |
Garzón-Roca et al. [108] | 0.67 | -5.28 | 8.850 | 2.455 | 36.4 | 5.436 | 6.849 | 182.9 | -32.61 |
Gumaste et al. [82] | 1.90 | 0.679 | 3.927 | 0.706 | 37.2 | 2.632 | 3.111 | 91.11 | 89.63 |
Christy et al. [112] | 0.75 | 0.303 | 1.489 | 0.258 | 34.6 | 1.663 | 2.564 | 31.54 | -25.26 |
Bennett et al. [73] | 0.86 | 0.337 | 2.175 | 0.375 | 43.7 | 1.390 | 1.942 | 33.16 | -14.169 |
Basha & Kaushik [72] | 0.64 | 0.287 | 0.991 | 0.155 | 24.4 | 2.124 | 3.169 | 54.25 | 15.681 |
Thamboo & Dhanasekar [70] | 1.13 | 0.423 | 3.083 | 0.551 | 48.8 | 1.871 | 3.093 | 39.12 | 12.89 |
Bröcker [113] | 1.26 | 0.422 | 2.713 | 0.502 | 39.8 | 1.434 | 2.037 | 40.95 | 26.08 |
Dymiotis & Gutlederer [114] | 1.02 | 0.433 | 2.453 | 0.425 | 41.7 | 1.374 | 2.057 | 31.89 | 2.026 |
Calderon et al. [71] | 3.10 | 0.647 | 9.942 | 1.943 | 62.7 | 5.487 | 6.395 | 212.1 | 210.08 |
Hendry & Malek [115] | 0.48 | 0.173 | 1.002 | 0.180 | 37.2 | 2.667 | 3.705 | 51.61 | -51.60 |
Dayaratnam [116] | 0.73 | 0.231 | 1.857 | 0.325 | 44.7 | 1.920 | 2.992 | 36.82 | -27.26 |
Veríssimo-Anacleton et al. [93] | 2.05 | 0.564 | 7.87 | 1.389 | 67.7 | 6.609 | 11.44 | 113.42 | 105.16 |
The values (*) represent the average, minimum and maximum values for (Fpred/Fexp), SD represents the standard deviation of the (Fpred/Fexp) ratio and Cov represents its coefficient of variation, MAD represents the mean absolute deviation, RMSE represents the root mean square error, MAPE represents the mean absolute percentage error and MPE represents the mean percentage error.
Table 6 presents the performance of the 18 empirical equations outlined in Table 5. It can be observed that, in terms of average (Fpred/Fexp), Eq. (3) and the equation proposed by Gumaste et al. [82] exhibit the best values for prism tests. They are closely followed by the equations developed by Thamboo & Dhanasekar [70] and Dayaratnam [116], with coefficients of variation of 0.498, 0.417, and 0.863, respectively. Notably, for the Dayaratnam [116] equation, it should be noted that Fig. 5 (n) reveals an underestimation of strength for fwv greater than 15 MPa. The equations proposed by Christy et al. [112] and Bennett et al. [73] also demonstrate average values that are relatively close to one. These equations exhibit particularly low MAPE and MPE indicators compared to the other equations examined. However, it should be noted that the Christy et al. [112] equation, as depicted in Fig. 5 (i), underestimates masonry strength when fwv exceeds 20 MPa
On the other hand, the equation developed by Hendry & Malek [115] significantly underestimates the compressive strength, with an average value (Fpred/Fexp) of 0.60. Consequently, it displays the lowest error index (MPE), as clearly illustrated in Fig. 5 (l), where a substantial number of points fall below the diagonal line.
In contrast, the equations proposed by Engesser [109], Calderon et al. [71], and Garzón-Roca et al. [108] exhibit a significant overestimation of masonry strength. These models have the highest average values (Fpred/Fexp) among the examined equations, as well as RMSE, MPE, and MAPE error indicators that are considerably higher than those of other models. This observation is evident in Fig. 5 (g), (h), and (r), respectively. Thaickavil & Thomas [69] also noted an overestimation of strength for the Engesser [109] and Garzón-Roca et al. [108] equations.
Regarding the wallette tests, Eq. (4) as well as the equations proposed by Bennett et al. [73], Dymiotis & Gutlederer [114], Kaushik et al. [83], and Thamboo & Dhanasekar [70] exhibit the best values in terms of average (Fpred/Fexp). Notably, the equation presented by Kaushik et al. [83], which originally had an average (Fpred/Fexp) of 1.38 for the prism tests, decreases and approaches to one in the wallette tests. In contrast, the equation proposed by Christy et al. [112] clearly underestimates the compressive strength for the wallette tests, despite performing well in the prism tests.
Moreover, the equations proposed by Lumatarna et al. [14] and Calderon et al. [71] exhibit a significant overestimation of compressive strength, which was observed in the prism tests but not in the wallette tests. Conversely, the equation by Garzón-Roca et al. [108] clearly underestimates the compressive strength in the wallette tests, while it overestimates the compressive strength in the prism tests
Among the collected equations, only Gumaste et al. [82] and Thamboo & Dhanasekar [70] proposed different equations to evaluate the compressive strength using prism and wallette tests. The equation proposed by Gumaste et al. [82] for prism tests demonstrates excellent results. However, the equation for wallette tests tends to overestimate the compressive strength, with a mean value (Fpred/Fexp) that is almost twice the expected value, along with relatively high statistical parameters (MPE, MAPE, and RMSE). On the other hand, the equations presented by Thamboo & Dhanasekar [70] yield satisfactory results for both types of tests, making them suitable for evaluating the compressive strength of masonry.
Padalu and Singh [86] evaluated the performance of empirical equations on Indian masonry and concluded that the equations proposed by Kaushik et al. [83], Dymiotis and Gutlederer [114], and Christy et al. [112] exhibited the best performance for masonry in India. The results of our present study closely align with these findings, as Padalu and Singh [86] also examined masonry samples with strengths below 15 MPa. It is evident that these equations yield satisfactory results for masonry with strengths below 15 MPa. Therefore, for this present study, Eq. (4) has been chosen to calculate the compressive strength of masonry.
2.4. Masonry Shear Strength
Regarding shear strength, numerous experimental tests did not provide this value. Therefore, equations have been proposed in the literature to establish a relationship between compressive strength and shear strength (fwu). In this present study, the equation proposed by Liberatore et al. [117] was selected to calculate the missing values.
$${f_{wu}}=0.285\sqrt {{f_{wv}}}$$
5