1 Introduction

A number of researchers such as Brace et al. (1966), Bieniawski (1967), Brady and Brown (1993), Martin and Chandler (1994), Hatzor and Palchik (1997), Pettitt et al. (1998), Eberhardt et al. (1999), Katz and Reches (2004) and Cai et al. (2004) have investigated different stages of stress–strain behavior of brittle rocks during uniaxial compression. Two stress–strain diagrams in Fig. 1 obtained by the author for carbonate rocks, exhibit three main characteristic stress levels (crack initiation stress σci, crack damage stress σcd and uniaxial compressive strength σc) which are represented by points A, B and C, respectively, over the total volumetric strain curve. The crack initiation stress (σci, point A) is the stress level at which microfracturing begins. The point A is the end of the elastic stage (linear portion) of stress–strain curve. The crack damage stress (σcd, point B) is the stress at the onset of dilation: when σcd is attained, the rock volume begins to increase (Schock et al. 1973; Brace 1978; Paterson 1978; Palchik and Hatzor 2002). The crack damage stress (σcd, point B) is the stress level at which maximum total volumetric strain (εcd) is attained. The uniaxial compressive strength (σc, point C) is the maximum axial stress (at failure). In Fig. 1, strain εcd at σcd (point B) is the maximum total volumetric strain, and strain εa max at σc is the maximum axial strain.

Fig. 1
figure 1

Stress–strain behavior of brittle rocks during uniaxial compression. Crack initiation stress (σci), crack damage stress (σcd) and uniaxial compressive strength (σc) are represented by points A, B and C, respectively. The total volumetric strain (εv) is calculated as a sum of the component strains: εv = εa + εR1 + εR2, where εa is axial strain, and εR1 and εR2 are radial strains measured in orthogonal directions (Palchik and Hatzor 2002): a type 1—total volumetric strain curve has a reversal point (B) and σcd < σc; b type 2—there is no reversal point in total volumetric strain curve, and σcd = σc

Martin and Chandler (1994), Eberhardt et al. (1999) and Palchik and Hatzor (2002) have shown that the crack damage stress (σcd) is defined as the point (see point B in Fig. 1a) where a total volumetric strain reversal occurs and unstable crack growth begins. In this case (see Fig. 1a), crack damage stress σcd is lower than the uniaxial compressive strength (σc). Indeed, Brace et al. (1966), Bieniawski (1967), Martin (1993), Pettitt et al. (1998), Eberhardt et al. (1999), Heo et al. (2001) and Katz and Reches (2004) have found that the crack damage stresses σcd of granites, sandstones and quartzite vary from 0.71σc to 0.84σc. They have also shown that the ratios σcic and σcicd for above-mentioned rocks range from 0.39 to 0.6 and from 0.52 to 0.82, respectively.

Hatzor and Palchik (1997) and Palchik and Hatzor (2002) have shown that there exist total volumetric strain curves (in heterogeneous carbonate rocks) which do not have any point of reversal, with the maximum total volumetric strain (εcd) attained at the uniaxial compressive strength (σc). In this case, crack damage stress is equal to uniaxial compressive strength (i.e. σcd = σc, B = C in Fig. 1b). Thus, there are two types of total volumetric strain curves in brittle rocks: type 1 (see Fig. 1a), with a point of reversal (B) in the total volumetric strain curve, and σcd < σc; and type 2 (see Fig. 1b), where the total volumetric strain curve has no reversal point and, therefore, σcd = σc.

Type 1 (Fig. 1a) curves have been studied by Brace et al. (1966), Bieniawski (1967), Pettitt et al. (1998), Eberhardt et al. (1999), Heo et al. (2001), Katz and Reches (2004), etc., whereas little attention has been paid to type 2 mechanical behavior of brittle rocks (Fig. 1b). In this paper, we intend to study relations between characteristic compressive stress levels, strains and mechanical properties of heterogeneous carbonate rocks exhibiting type 2 behavior of the total volumetric strain curve (i.e. at σcd = σc).

2 Testing and test results

Mechanical properties of carbonate rock samples exhibiting type 2 behavior of the total volumetric strain curve are summarized in Table 1. These samples were collected from Adulam chalk, Aminadav dolomite, Bina limestone, Yarka limestone, Yagur dolomite, and Nekorot limestone formations. The NX (d = 54 mm) sized cylindrical rock samples having the ratio L/d = 2 (here L and d are the length and diameter of a sample, respectively) were prepared. The samples were ground to the planeness of 0.01 mm and cylinder perpendicularity within 0.05 radians. Prior to testing, rock samples were oven dried at the temperature of 110°C for 24 h.

Table 1 Observed values of compressive stress levels, elastic modulus, dry bulk density, Poisson’s ratio, porosity, maximum axial strain and maximum total volumetric strain

Uniaxial compressive tests were performed at the Rock Mechanics Laboratory of the Ben-Gurion University. The tests were conducted using a load frame (TerraTek system, model FX-S-33090) at a constant strain rate of 10−5/s. The load frame operates under a hydraulic closed-loop servo-control. The load frame stiffness and maximum axial force are 5 × 109 N/m and 1.4 MN, respectively. The axial strain cantilever set has a 10% strain range, and the radial strain cantilevers have a strain range limit of 7%, with the linearity of 1% over the full scale in both sets. The servo-controlled press description, physical and mechanical properties of the studied rock formations are presented in detail elsewhere (Hatzor and Palchik 1997; Palchik and Hatzor 2002, 2004).

The results of uniaxial tests are given in Table 1. The latter presents the values of crack initiation stress (σci), σcd = σc (here σcd is the crack damage stress and σc is the uniaxial compressive strength), elastic modulus (E), dry bulk density (ρ), Poisson’s ratio (ν), porosity (n), maximum axial strain εa max (at σc) and maximum total volumetric strain εcd (at σcd) for each of 24 studied rock samples.

Porosity (n, %) was calculated from the measured values of dry bulk density (ρ in Table 1) and specific gravity of the solids (G s = 2.7–2.8 g/cm3): n = [1 − (ρ/G s)] × 100%. The precision of the porosity estimation is 0.1%. The elastic modulus (E) and Poisson’s ratio (ν) were calculated using linear regressions along the linear portion (elastic stage) of the stress–strain curve. The total volumetric strains (εv) calculated as a sum of the component strains (εa, εR1 and εR2) were plotted versus axial stresses for each of 24 studied rock samples. Figure 2 demonstrates examples of axial stress–total volumetric strain curves for five rock samples (ad 80, ca3541, gn3-2a, gn2-1b and rc9) exhibiting εcd = 0.134% at σcd = σc = 174 MPa, εcd = 0.149% at σcd = σc = 174 MPa, εcd = 0.161% at σcd = σc = 150 MPa, εcd = 0.236% at σcd = σc = 177 MPa and εcd = 0.134% at σcd = σc = 63.1 MPa, respectively, in point the B = C.

Fig. 2
figure 2

Examples of observed axial stress–total volumetric strain curves (type 2) for five studied rock samples

In Table 1, the observed values of crack initiation stress (σci) and crack damage stress or uniaxial compressive strength (σcd = σc) vary from 15 to 145 MPa with the mean of 61.7 MPa and from 31.9 to 273.9 MPa with the mean of 101.4 MPa, respectively. The difference (D) between σc and σci and the ratio k = σcic for each of the studied rock samples were calculated. Values of D and k range from 11.2 to 128.9 MPa (D mean = 39.7 MPa) and from 0.45 to 0.78 (k mean = 0.62), respectively, for all studied samples. Here, standard deviations (Δ) of the mean D and k values are significant—29.2 MPa and 0.09, respectively. Standard deviation of the mean has been calculated as follows:

$$ \Updelta = \sqrt {\frac{{\sum\nolimits_{i = 1}^{n} {(q_{i} - q_{m} )^{2} } }}{n - 1}} $$
(1)

where i = 1, 2,…, n is the number of observed sample (n = 24), q i is the value of the observed parameter (D or k) in the ith sample, q m is the arithmetic mean of parameters observed in n samples.

Note also that the values of D and k are not constant even for a single set of samples within the same rock formation: 11.9 MPa < D < 29.4 MPa (0.47 < k < 0.78), 72.2 MPa < D < 128.9 MPa (0.53 < k < 0.59), 12.8 MPa < D < 49.8 MPa (0.47 < k < 0.63), 20.5 MPa < D < 63.9 MPa (0.6 < k < 0.66) and 40.5 MPa < D < 97 MPa (0.45 < k < 0.75) for Adulam chalk, Aminadav dolomite, Bina limestone, Yagur dolomite and Nekorot limestone, respectively. The standard deviation (Δ) of the mean k within the same rock formation ranges from 0.03 to 0.12. For example, Nekorot limestone has a large Δk = 0.12, whereas Yagur dolomite exhibits relatively a small Δk = 0.03.

Table 1 demonstrates the values of elastic modulus (E), porosity (n) and Poisson’s ratio (ν) for each of 24 studied rock samples. Here, 6,100 MPa < E < 64,000 MPa, 5.4% < n < 28.5% and 0.13 < ν < 0.31. Mean values of E, n and ν are 31,600 MPa, 16.5% and 0.23, respectively, and standard deviations (Δ) of mean E, n and ν values are 17,300 MPa, 6.5% and 0.04, respectively. The values of E, n and ν in heterogeneous carbonate rocks are also non-constant even for samples within the same rock formation. For example, Bina limestone exhibits 21,000 MPa < E < 42,830 MPa, 7.5% < n < 26.4% and 0.2 < ν < 0.31.

In Table 1, maximum total volumetric strain (εcd) and maximum axial strain (εa max) range from 0.084 to 0.9% with the mean of 0.2% and from 0.19 to 1.36% with the mean of 0.39%, respectively. Thus, the maximum axial strain (εa max) is 1.5–2.5 times larger than the maximum total volumetric strain (εcd) in case where crack damage stress (σcd) is equal to uniaxial compressive strength (σc). For example, rock sample bina 7 exhibiting εcd = 0.132% at σcd = 64.2 MPa and εa max = 0.29% at σc = 64.2 MPa (see Fig. 1b), has the ratio εa maxcd = 0.29/0.132 = 2.2.

3 Relations between mechanical properties and compressive stresses

3.1 Effect of elastic modulus, porosity and E/n ratio on σci and σcd = σc

Figure 3a shows how the elastic modulus (E) influences the values of σci and σcd = σc. Relations σci − E and σc − E best follow a polynomial law with good squared regression coefficients R 2 = 0.86 and 0.91 for σci − E and σc − E, respectively. An increase in the elastic modulus (E) from 6,100 to 64,000 MPa leads to an increase in σci and σc = σc values from 15 to 145 MPa and from 31.9 to 273.9 MPa, respectively. Hence, an increase in the elastic modulus by a factor of 10.5 leads to an increase in the values of σci and σcd = σc by a factor of 9.7 and 8.6, respectively.

Fig. 3
figure 3

Relations between compressive stress levels and a elastic modulus (E), and b porosity (n)

On the other hand, values of σci decrease from 145 to 15 MPa and from 273.9 to 31.9 MPa, respectively, with porosity increase (n, Fig. 3b) from 5.4 to 28.5%. In Fig. 3b, polynomial correlations (R 2 = 0.76–0.79) between the porosity (n) and σci and σcd = σci values are obtained.

An increase in the values of stresses σci and σcd = σc with increasing ratio E/n is shown in Fig. 4a. The latter demonstrates that σci and σcd = σc are well correlated (R 2 = 0.82–0.86) with the ratio E/n. It is not surprising, since the compressive stresses σci and σcd = σc depend on the elastic modulus E (Fig. 3a) and porosity n (Fig. 3b).

Fig. 4
figure 4

Relations between compressive stress levels and ratios a E/n and b E

3.2 Effect of E/λ ratio on σci and σcd = σc

Porosity n is a measure of void space (pores and open cracks) and represents a ratio between the void space (V p) and bulk volume (V).

$$ n = \frac{{V_{\text{p}} }}{V} $$
(2)

The bulk volume (V) is the initial volume of a sample before loading, and therefore, it does not reflect the change in the volume due to compression. Palchik and Hatzor (2002) have proposed to use the ratio between the volume of voids and change in the bulk volume due to compression, since such ratio reflects the mechanical behavior of the rock matrix. This ratio can be represented as a ratio between the volume of voids (V p) and the maximum compaction (V c) of a rock sample:

$$ \lambda = \frac{{V_{\text{p}} }}{{V_{\text{c}} }} $$
(3)

where V c is the maximum decrease in a sample volume (maximum compaction of a sample), which is attained at the maximum total volumetric strain εcd (at crack damage stress σcd).

The parameter λ can be presented as

$$ \lambda = \frac{{V_{\text{p}} }}{{V_{\text{c}} }} = \frac{{V_{\text{p}} }}{V}/\frac{{V_{\text{c}} }}{V} = \frac{n}{{\varepsilon_{\text{cd}} }} $$
(4)

where V c/V = εcd, εcd is the maximum total volumetric strain at the crack damage stress σcd.

When we use λ = V p/V c instead of n = V p/V, the ratio E/n can be rewritten as E/λ. Relations between E/λ and compressive stress levels are presented in Fig. 4b. From Fig. 4b it is clear that the values of σcd = σc and D = σc − σci are well correlated with E/λ. Here, power dependences between σcd = σc (R 2 = 0.96), D (R 2 = 0.87) and E/λ with good squared regressions coefficients R 2 are obtained:

$$ \sigma_{\text{cd}} = \sigma_{\text{c}} = a\left(\frac{E}{\lambda }\right)^{b} $$
(5)
$$ D = c\left(\frac{E}{\lambda }\right)^{d} $$
(6)

where coefficients a = 2.75, b = 0.6, c = 0.82 and d = 0.64.

Note that the use of E/λ ratio (Fig. 4b) instead of E/n (Fig. 4a) versus the compressive stress σcd = σc allows us to increase the value of R 2 from 0.86 to 0.96. Hence, the effect of the ratio E/λ on uniaxial compressive strength (σc) is more pronounced than the effect of E/n.

4 Conclusions

The mechanical behavior of heterogeneous carbonate rocks exhibiting total volumetric strain curves of type 2 was studied. Studied rock samples exhibiting a wide range of mechanical properties (31. 9 MPa < σcd = σc < 273.9 MPa, 6,100 MPa < E < 64,000 MPa and 5.4% < n < 28.5%) were collected from different geological settings of Israel. From the results of this study it can be concluded that:

  • Crack initiation (at the crack initiation stress σci) for studied heterogeneous carbonate rocks occurs at 0.45σc/0.78σc (at significant standard deviations ΔD = 29.2 MPa and Δk = 0.09 for all studied samples).

  • Values of the difference (D) between the uniaxial compressive strength and crack initiation stress, and the ratio (k) between the crack initiation stress and uniaxial compressive strength are not constant even for samples within the same rock formation. The standard deviation (Δ) of the mean k within the same rock formation varies between 0.03 and 0.12.

  • Values of the maximum axial strain (εa max) and maximum volumetric strain (εcd) at σcd = σc are 0.19/1.36% and 0.084/0.9%, respectively. At σcd = σc, the maximum axial strain (εa max) is 1.5–2.5 times the maximum volumetric strain (εcd).

  • The ratio between the elastic modulus (E) and parameter λ strongly influences the values of σcd = σc and D = σc − σci. Power dependencies between σcd = σc, σc − σci and E/λ are obtained. Parameter λ is a ratio between the volume of voids (V p) and the maximum compaction (V c) of rock sample. Parameter λ is calculated as ncd, where εcd is the maximum total volumetric strain.