**1. Introduction**

Concrete centrifugation is the process of forming and compacting concrete mix due to the normal (radial) force generated during the spinning of the mold around its longitudinal axis at a speed of 500–700 rpm. Due to its specificity, this method is applicable only to hollow elements. As a result of the centrifugation process, concrete with a heterogeneous and layered structure is obtained [1–3]. Thus, the structure is different from that of precast or monolithic concrete elements, which in practice can be considered homogeneous. The spun concrete is marked by the fact that components with a larger mass (coarse grains) come to the outside of the cross-section, while components with a smaller mass (cement slurry) tend to remain inside (Figure 1). Under these conditions, the outer layer can achieve high compactness and, after hardening, high strength and resistance to chemical and mechanical impacts. The inner layer, on the other hand, consists of highly compacted, very thick cement paste and may, after hardening, obtain special resistance to water permeability.

**Figure 1.** The structure of spun concrete in the wall thickness of a power pole.

One of the first to describe the structure of spun concrete used in pipe production was Marquardt [4]. Based on his observations, he found that a greater di fference between the specific gravity of the concrete mixture components led to faster and deeper fractionation. He recommended that well-sorted, mixed aggregate grains with similar petrographic properties and a maximum diameter of 15 mm were preferred for concrete mixtures subjected to centrifugation. He also discussed the change in cement content as a function of the wall thickness of spun concrete elements. He found that cement at 83% of the wall thickness is quite evenly distributed, and an increase in its content is observed only in the inner layer of the section with a thickness of 2–5 mm. He proposed to use layered centrifugation at lower mold rotational speeds to reduce layering.

In [5], Kuranovas and Kvedaras presented the concrete centrifugation process as a four-phase process. The first phase is feeding the concrete mix inside the mold, with its even distribution over the length of the mold at a low centrifuge rotational speed. Afterward, the rotational speed of the mold increases, and the process enters the second phase, where the centrifugal force begins to compress the concrete mix and a layered wall of the concrete section is formed. The third phase is a further increase in the rotational speed of the mold, at which the wall thickness of the concrete section is compacted and stabilized with some of the mixing water squeezed inside. The fourth phase occurs at the maximum rotational speed of the mold, during which the concrete mix is further compacted and excess mixing water is squeezed out. The element's wall reaches its designed thickness.

Kliukas, Jaras, and Lukoševiˇciene in [ ˙ 6] found that each phase of the concrete centrifugation process has a specific rotational speed, which depends on the dimensions of the element and the time needed to obtain the selected speed. According to [6], the duration of the first phase should be 3–4 min, the second and third phase should be 1–2 min, and the fourth phase should be 10–15 min. It should be noted that the centrifugation time of an element in a steel form also depends on the initial water–cement ratio of the concrete mix and the pressure caused by centrifugal force, depending on the rotational speed and the diameter of the product.

Adesiyun et al. [7–9] used a computer image analysis method to describe the structure of spun concrete allowing extraction of quantitative and qualitative information contained in the image of an element wall cross-section. For a precise description of the structure, the wall thickness of the sample was divided into 20 strips with a width of 2.25–3.00 mm (depending on the overall wall thickness). The tests were carried out with di fferent combinations of concrete mix parameters, such as sand point (25–50%), cement content (410–530 kg/m3), amount of plasticizer (0–2%), spinning speed (400–700 rpm), and its time (5–11 min). Computer image analysis allowed for graphical representation of the aggregate, cement matrix, and air-void distribution as a function of the wall thickness in individual samples. It was found that the volume fraction of aggregate decreases from the outside of the centrifuged sample to the inside, while the cement matrix content changes in the opposite direction (in the inner zone, its content is almost 100%). The air content in the cross-section of the wall was higher for internal rather than external layers. Segregation of components was found in all tested samples.

Völgyi et al. [10,11] found that segregation of the concrete mix during centrifugation depends to a large extent on the excess of the cement paste and the degree of compaction. Properties such as strength, porosity, hardness, and composition vary across the wall of the spun concrete due to the segregation of components. The latter can be reduced by using less paste and optimal compaction energy. A relationship between the consistency and the advisable compaction energy was proposed. The formula for the optimal centrifuge settings, i.e., time and speed of spinning, was derived to obtain the best strength properties of the sample.

This paper focuses on methods of assessing the internal structure of spun concrete, which can be used in practice to verify the selection of spinning parameters. In the first part of the work, standard methods for determining the composition of hardened concrete were used to assess its variability across the element wall. Next, based on preliminary results presented in [12], imaging techniques followed by appropriate image analysis were utilized to determine spatial distribution of pores, aggregate, and cement matrix. Comparison of obtained results was carried out for the samples of spun concrete power pole E10.5/6c made by one of the Polish manufacturers of prestressed concrete E-poles. The capabilities and performance of the research methods used were discussed. Conclusions regarding their potential use in practice were formulated.

### **2. Preparation of Test Samples**

The E10.5/6c pole used for testing was made in one of the Polish prefabrication plants and was a precast, partly prestressed element shaped as a truncated cone with hollow core. The length of the pole was equal to *L* = 10.5 m. The annular shape of the cross-section resulted from the adopted technology of manufacturing poles by the method of concrete centrifugation in longitudinally unopenable steel mold [13]. The outer diameter of the E10.5/6c column increased from *dt* = 173 mm at the top to *db* = 330.5 mm at the base with a constant taper *t* = 15 mm/m. The equivalent characteristic peak resistance force, applied at a distance of 0.17 m from the top, was *Pk* = 6 kN. The E10.5/6c pole is designed to be produced from C40/50 spun concrete.

When designing power poles made of spun concrete, it was taken into account that they are exposed to direct environmental impacts, described by standard [14], class XC4 (cyclically wet and dry concrete surfaces exposed to contact with water), and for structures located near motorways, class XF2 (vertical concrete surfaces exposed to freezing and de-icing agents from air). The durability of spun prestressed concrete poles is ensured by appropriate thickness of the concrete cover of the reinforcement. Concrete that protects steel against corrosion is marked by limit values describing, among others, the quantity and quality of components (e.g., minimum cement content, water–cement ratio *<sup>w</sup>*/*<sup>c</sup>*, and classified aggregates), as well as the minimum compressive strength of concrete and its absorbability. For the XC4 and XF2 environmental exposure class, the standard [14] imposed a minimum cement content of 300 kg/m3, maximum *w*/*c* = 0.5, and minimum concrete strength class C30/37. A concrete mix recipe was obtained from the manufacturer for the production of the E10.5/6c pole (Table 1).


**Table 1.** Recipe for concrete mix used in the production of E10.5/6c pole.

For laboratory tests, from an E10.5/6c power pole, about a 200-mm-long fragment was cut off (Figure 2a) to obtain a sample in the shape of a hollow truncated cone. The wall thickness of the fragment was about 60 mm. The fragment of the pole was then cut along the generatrix into slices about 10–20 mm thick (Figure 2b), with the prestressing wires being avoided. One of the slices was intended for testing the concrete structure with the use of imaging techniques, while the other was intended for testing the composition of hardened concrete using laboratory methods.

**Figure 2.** Preparation of test samples: (**a**) cutting off the top of the pole; (**b**) a slice of concrete intended for laboratory testing.

### **3. Determining the Composition of Hardened Concrete**

For testing the composition of hardened concrete, an estimation method was used in accordance with the instructions [15] based on the determination of apparent density of concrete, the content of parts insoluble in HCl, and the amount of components attached to the binder during its hardening. Since there is no direct method for determining the cement content of a concrete sample, the procedure consists of determining the content of soluble silica and calcium oxide and, on this basis, calculating the cement content. The method is based on the fact that the silicates in Portland cement are much more soluble than the silicate components normally contained in the aggregate. The same applies to the relative solubility of the calcium oxide components in cement and aggregate (except, however, limestone aggregate). The methods for determining the Portland cement content in concrete are described in standards [16,17], as well as in the already mentioned manual [15]. In addition, the extension of the research to determine water absorption and specific density according to [18,19] made it possible to calculate the total and open porosity. The composition of hardened concrete was determined independently for four layers of concrete pole section as shown in Figure 2b.

Due to the layered structure of spun concrete, for laboratory tests, samples were prepared by longitudinally cutting the concrete slices (Figure 2b) into four parts (as shown in Figure 3). Samples (layers) were marked with consecutive numbers 1–4, where sample 1 describes the inner layer of the annular cross-section, and sample 4 describes the outer layer. In addition, sample 5 was an entire slice of the concrete wall without splitting into layers. For aggregate with a grain diameter of up to 40 mm, the weight of the concrete sample should be at least 3 kg according to [15]. During the tests, due to the dimensions of separated centrifuged concrete samples, this condition was not met. Moreover, while the concrete slices were being cut, it was not possible to ge<sup>t</sup> distinct layers of precise dimensions (as in Figure 1); thus, the boundaries between the layers were in practice inexact, which was especially visible for the inner layer No. 1.

**Figure 3.** Sample shapes for laboratory tests.

The apparent density of concrete ρ*c* was determined via the hydrostatic method due to the irregular shape of the samples. Determination was performed according to [18]. The samples were first dried at 105 ◦C to constant weight, i.e., until the changes in mass during 24 h were less than 0.2%. Dried samples were weighed after that. Then, the samples were saturated by immersion in water at a temperature of 20 ◦C until the changes in mass during 24 h were less than 0.2% and weighed. The volume of each sample was determined on the basis of its apparent mass in water and mass in air according to the following formula:

$$V = \frac{m\_d - m\_w}{\rho\_w} \tag{1}$$

where *ma* is the mass in air of saturated sample, *mw* is the apparent mass in water of saturated sample, and ρ*w* = 998 kg/m<sup>3</sup> is the density of water at 20 ◦C. When weighing the samples in water, the correction due to apparent mass of suspension wire and holder was made. The apparent density of dry concrete was calculated as

$$
\rho\_{\mathcal{E}} = \frac{m\_{\mathcal{O}}}{V} \tag{2}
$$

where *mo* is the mass of dry sample. At the same time, water absorption of concrete can be calculated as

$$m\_{\rm uv} = \frac{m\_{\rm a} - m\_{\rm o}}{m\_{\rm o}} \tag{3}$$

The procedure for determining the content of parts insoluble in HCl was as follows: after comminution, sieving through a 1-mm sieve, and drying to constant weight at 105 ◦C, the samples were sieved through a 0.2-mm sieve. Samples of about 2 g were tested using an aqueous solution (1:3) of hydrochloric acid (HCl) to determine the content of insoluble parts. Each sample was placed in a 250-mL beaker and 100 mL of HCl solution was added at room temperature. The contents of the beaker were mixed and the insoluble residue was triturated with a glass rod. After 15 min, the supernatant fluid was decanted. Next, 50 mL of HCl solution was poured onto the precipitate in a beaker and placed in a water bath at 90 ◦C for 15 min. Then, the contents of the beaker were washed twice with hot water and decanted. The precipitate remaining in the beaker was flooded with 50 mL of 5% Na2CO3 and placed in a water bath at 90 ◦C for 15 min, before being washed twice with hot water and decanted. The remaining precipitate was flooded with 50 mL of water, acidified with HCl (1:3) against methyl orange, while adding an excess of 3–4 drops of acid to the neutralized solution, after which the contents of the beaker were filtered. The filtered precipitate was washed six times with hot water until the reaction due to chlorides ceased. The washed precipitate was transferred to a weighed porcelain crucible and, after burning, the filter was calcined at 1000 ◦C to finally determine the mass of parts insoluble in HCl. Typically, it is assumed that the aggregate mass in the tested sample is equal to the mass of parts insoluble in HCl determined as a result of the analysis. However, when determining the aggregate content in this study, a preliminary analysis was performed to determine the aggregate

behavior under the influence of hydrochloric acid. It turned out that basalt in the form of aggregate is resistant to HCl acid, while, after being pulverized, it dissolved partly in hydrochloric acid. Probably weathered basalt was used for concrete in some parts of the aggregate 8–11 mm (Table 1) [20]. Thus, when determining the aggregate content, account was taken of the presence of HCl-soluble parts in the aggregate, by introducing the appropriate correction. Finally, the aggregate mass percentage was calculated as

$$\mathbb{C}\_{4\% \text{S}} \text{\%} = \mathbb{C}\_{\text{ins} \text{\%} \text{\%}} \cdot \text{corr} = \frac{m\_{\text{ins}}}{m\_0} \cdot 100\% \cdot \text{corr} \tag{4}$$

where *Cins,*% is the mass percentage of insoluble parts in concrete, *mins* is the mass of insoluble parts remaining from the sample of concrete, *mo* is the initial mass of dry concrete sample, and *corr* =1/*Cins,agg,%* is the correction factor, in which *Cins,agg,%* stands for mass percentage of HCl-insoluble parts in the aggregate. The aggregate content in kg/m<sup>3</sup> was calculated as

$$\mathcal{C}\_{\text{a}\%} = \mathcal{C}\_{\text{a}\%\%} \cdot \rho\_c \tag{5}$$

The content of compounds attached to cement during the setting and hardening of concrete (which are H2O and CO2) was determined on the basis of calcining losses. The samples were prepared according to the procedure for determining the aggregate content in concrete; they were additionally ground on a 0.06-mm sieve. Calcining losses were determined in an oxidizing atmosphere, while the samples were heated in the air at 950 ◦C until a constant sample mass was obtained. The constant mass was determined by successive heating for 15 min, followed by cooling and weighing. The percentage calcining loss was determined by the following formula:

$$\mathcal{C}\_{\text{att,\%}} = \frac{(m\_o - m\_{\text{cal}})}{m\_o} \cdot 100\% \tag{6}$$

where *mo* is the the mass of the sample tested, and *mcal* is the the mass of the calcined sample. The obtained value of *Catt*,% was converted to the content of attached ingredients expressed in kg/m<sup>3</sup> according to the following formula:

$$\mathcal{C}\_{att} = \frac{\mathcal{C}\_{att, \%}}{100\%} \cdot \rho\_c \tag{7}$$

Finally, the content of cement *Ccem* was calculated from the following condition:

$$\mathcal{C}\_{\text{ccm}} = \rho\_{\text{c}} - \mathcal{C}\_{\text{a}\%\text{'}} - \mathcal{C}\_{\text{att}} \tag{8}$$

The density of concrete ρ*cr*, defined as the ratio of the mass of the dried sample to the volume of the solid part (without taking into account the volume of voids), was determined using the Le Chatelier flask method [19]. On the basis of the values of ρ*c* and ρ*cr*, total porosity was calculated as

$$\phi\_{p,tot} = 1 - \frac{\rho\_c}{\rho\_{cr}}\tag{9}$$

Open porosity φ*p,op* was estimated based on water absorption *nw* as follows:

$$
\phi\_{p,op} = n\_w \frac{\rho\_c}{\rho\_w} \tag{10}
$$

The obtained parameters of concrete in the particular layers are presented in Table 2.


**Table 2.** Test results and calculations for individual samples 1–4 and average sample 5 (Figure 1).

### **4. Assessment of Morphology of Concrete Microstructure**
