Base Material (BM). The material used in this project was Alhambra Formation (20), which is composed of sand, gravel, and pebbles in a reddish loamy clay matrix. This accounts for 90% of the dry ingredients. This Formation is used in many of the surviving rammed earth constructions in the city of Granada (Spain), including fortifications, dwellings and walls. Table 1a shows the physical and chemical parameters of the BM obtained from tests described in the UNE standard (21–30). Table 1b shows the physical and mechanical parameters obtained through the UNE-EN standard (31), articles and books (32–34) and the UNE standard (35–36).

Figure 1 shows the granulometric characteristics of the BM and Houben and Doat's (37) grading envelope for soil used in rammed earth constructions (<20% clay and >45% sand). Our BM soil meets these requirements, at least up to silt size (0.06 mm).

The BM is very heterogeneous in texture, a factor that affects not only the coarse but also the fine fraction, silt and clay, which comprise between 22% to 38% of the soil, although this does not affect the suitability of the material, as shown in the grading envelope in Figure 1.

Under X-ray diffraction crystallography, at <50 µ the soil is mainly composed of phyllosilicates, calcite, dolomite, portlandite and quartz. In terms of mechanical properties, BM has a Standard Proctor (SP) density of 2.06 g/cm³ (35) and a simple compressive strength (36) of ∼3.7 MPa (using PS-type samples with typical optimal humidity of 10%).

Hydrated lime. CL90-S content is 8% of soil weight (38).

White builder's cement (BL22.5X (39)): in this case, content is 2% of soil weight. This binder gives the sprayed BM initial mechanical strength without obstructing air permeability. The aim is not to obtain greater initial adherence to the receiving surface, since this is achieved by the projection pressure and is sufficient to ensure short-term durability.

Additives (40): two types of additives have been used. The first is a superplasticiser to reduce the amount of water needed for spraying. This is a versatile, colourless, high performance additive usually added to cement to increase fluidity without the need for additional water, thereby avoiding shrinkage and loss of mechanical strength. Secondly, after projection, a water repellent is applied to the earth to improve durability. This additive does not waterproof the surface of the sprayed earth, since this would affect transpiration.

The water used meets standard EHE-08 (41) specifications for the manufacture of cement.

Aggregate: this material is needed to improve the gradation of mixtures in which the base material contains over 30% of fines (clay and clayey-silt), to bring this proportion down to 20–25%. The aggregate is white in order to blend in with the structure, and must meet all EHE-08 (41) requirements, or other standards in place worldwide. The aggregate should be limestone or non-reactive silica, although other types can also be used provided they are laboratory-tested to ensure suitability. In the case reported here, 25% of the BM was replaced by 4–8 mm dolomite aggregate.

When a graph is plotted with aggregate particle size, Fuller's ideal grading curve (42) maximum size, Tmax 8, the BM grading envelope (tmax 10) and the resulting BM+ aggregate mixture curve for this size, it can be seen that the 4–8 aggregate fraction reduces the deviation from Fuller's ideal curve until the mixture curve almost matches the latter up to 2 mm sieve size and also significantly reduces the <2 mm particle content. This means that less water is needed to achieve PS density, which remains within approximately the same range (Figure 2).

The flow chart in Figure 3 shows both the field and laboratory methodology followed in this study.

A predominantly sandy soil with a certain amount of gravel, grit and fines with low to medium plasticity (22–24) (CL and CL-ML) was used, including around 22%–38% of <0.063 mm fraction. Such variations are to be expected when working with this type of granular geological formations classified under the Unified System of Soil Classification (46) as GC, SC, SC-SM and GC-GM. In the case of other formations, given the good results obtained when sprayed, suitable corrections should be made to adjust the mixture to the grading envelopes obtained here, which are in turn very close to those recommended by Houben and Doat (37).

Authors such as Rodríguez Ortíz (47) recommend the use of low expansion potential soils to prevent the appearance of shrinkage fissures in the sprayed earth. Our BM meets this specification; using the Lambe Determination of Expansivity in a Soil test, potential expansion should be “not critical” (25).

The density and compressive strength parameters were based on corresponding values obtained in existing walls built in Granada using this type of material and rammed earth techniques (dry density between 1.7–1.75 g/cm³ and wet density between 1.8–2.0 g/cm³; the real density of the soil should be around 2.6 g/cm³). These parameters have been shown to be particularly durable even after several hundred years.

Using SP energy, density increases significantly with respect to the tamping foot compaction technique, with a dry bulk density [1] of around 2.0 g/cm³ and a wet density (with around 6% humidity) [2] of around 2.2 g/cm³; logically, real density (G) is the same as indicated above, and compressive strength tested on oven-dried samples is around 3.5 MPa. The MP test was not included in the protocol because the densities and strength obtained after spraying only approach SP values. γ d = G 1 + e γ d ( 1 + W )

These values are associated with an “e” void [3] ratio of <0.3 (<0.4). This, together with dry bulk density (>1.85 g/cm³), among other parameters, allowed us to assign the SP compaction a “very dense” relative compaction value (33). With regard to the “S_r” saturation level [4], our level of 52% was acceptable since it allowed us to achieve SP density without sacrificing air permeability in compacted soil. The S_r values are consistent with the fact that the void ratio continues to decrease as compaction energy, and therefore density, increases, which is solely due to the air being expelled or compressed and dissolved in the water. This would not have been the case if the S_r value of the SP had been above 80%–85%, since air permeability is greatly reduced (34). These S_r values and void ratio show that even greater densities could be achieved by increasing compaction energy even above modified proctor (MP) levels. It is safe to say, therefore, that the process, at least in the case of these soils, does not end with MP energy. e = n 1 - n S r = W W s a t

To define the best blend using dolomite aggregate, different lime and cement dosages were studied in detail, and forced carbonation was tested in a CO₂ chamber. All tests were conducted using the same protocol and samples manufactured with SP energy (Table 2).

In summary, Table 3 shows the most important physical and mechanical parameters considered in samples manufactured using different blends both before and after spraying. The first are manufacturing using SP energy and the second by extracting 5 cm cores using a diamond bit cement sampler (48).

Several spray tests were needed to determine the best system (dry or wet mix) and spray gun according to the type and scope of the restoration or construction project.

The dry mix system is not recommended, since this causes significant segregation as the mixture is projected from the nozzle. When the mixture was sprayed, a large quantity of aggregate was separated from the soil by the pressure of the compressed air, and subsequently lost; it was impossible to determine the exact aggregate content of the sprayed mixture. We also noted a slight delay between projection of the dry mass and the water, making it impossible to control the volume of water added and to add a water reducing agent (superplasticiser).

The wet mix spray gun, meanwhile, had several advantages, including control of (i) added water, (ii) superplasticiser, (iii) and binders. This gave an evenly blended, dense and strong spray mixture while reducing rebound and poor adhesion.

Samples were taken of the sprayed mixture from different guns: DM, WM and WMW and the density and compressive strength were compared with the same results from the SP compaction tests using the same dosages (Table 4).

The mechanical strength of the mixture sprayed from the smallest gun (WMW), once totally dry, ranges from 3.8 to 4.2 MPa. This will increase over time due to the effect of natural carbonation. In the case of spray guns typically used in civil engineering works (WM), strength ranges from 4.3 to 4.8 MPa.

Spray tests were also crucial in establishing the order in which the ingredients are added to the mixture, since Projected Earth does not behave in the same way as other mortars mixed before spraying. Based on the test results, the following was determined to be the best order: first, all the water content, then all the cement + lime, half the soil, half the superplasticiser, half the soil and finally half the superplasticiser. If anti-shrinkage agents are used, these should be added at the end of the foregoing stages.

Figure 5 shows the density and strength of the earth samples plotted on Houben's (15) graph for other earth-based construction systems. It can be seen that density and strength are similar to that of rammed earth walls, which has from the outset been one of the aims of this project.

Figure 6 shows the practical application of the protocol described in this manuscript. In this case, projected earth was used to restore a rammed earth wall built using the calicastrado technique, in which a layer of lime slurry is applied between each block of earth. Seven years later, we can safely attest to the excellent durability of the restoration, even without the application of a layer of water-repellent. The soil used was the same as that used in the original construction, ensuring compatibility in terms of texture and colour.