Injection 3D Concrete Printing (I3DCP): Basic Principles and Case Studies

materials

Article Injection 3D Concrete Printing (I3DCP): Basic Principles and Case Studies

Norman Hack 1,* , Inka Dressler 2 , Leon Brohmann 1, Stefan Gantner 1, Dirk Lowke 2 and Harald Kloft 1

1 Institute of Structural Design, Technische Universität Braunschweig, Pockelsstr. 4, 38104 Braunschweig, Germany; [email protected] (L.B.); [email protected] (S.G.); [email protected] (H.K.) 2 Institute of Building Materials, Concrete Construction and Fire Safety, Technische Universität Braunschweig, Beethovenstr. 52, 38106 Braunschweig, Germany; [email protected] (I.D.); [email protected] (D.L.) * Correspondence: [email protected]

Received: 25 January 2020; Accepted: 27 February 2020; Published: 1 March 2020

Abstract: Today, the majority of research in 3D concrete printing focuses on one of the three methods: firstly, material extrusion; secondly, particle-bed binding; and thirdly, material jetting. Common to all these technologies is that the material is applied in horizontal layers. In this paper, a novel 3D concrete printing technology is presented which challenges this principle: the so-called Injection 3D Concrete Printing (I3DCP) technology is based on the concept that a fluid material (M1) is robotically injected into a material (M2) with specific rheological properties, causing material M1 to maintain a stable position within material M2. Different to the layered deposition of horizontal strands, intricate concrete structures can be created through printing spatially free trajectories, that are unconstrained by gravitational forces during printing. In this paper, three versions of this method were investigated, described, and evaluated for their potential in construction: A) injecting a fine grain concrete into a non-hardening suspension; B) injecting a non-hardening suspension into a fine grain concrete; and C) injecting a fine grain concrete with specific properties into a fine grain concrete with different properties. In an interdisciplinary research approach, various material combinations were developed and validated through physical experiments. For each of the three versions, first architectural applications were developed and functional prototypes were fabricated. These initial results confirmed both the technological and economic feasibility of the I3DCP process, and demonstrate the potential to further expand the scope of this novel technology.

Keywords: Injection 3D Concrete Printing; digital concrete; concrete 3D printing

1. Introduction While additive manufacturing (AM) is already a well-established technology and integral part of the production in many industrial sectors, the potential of AM to also become a key technology in the construction industry is clearly evident today. Compared to traditional concrete construction, AM offers a number of advantages: through the use of computer-controlled machines, building components can be mass-customized and produced individually in series, thus combining the benefits of traditional craftsmanship and industrialized production. In addition, digital control of 3D printing hardware allows the geometric complexity of components to be increased in order to improve component efficiency, for example by means of structural optimization. As such, material can be placed exclusively where it is structurally required. A further advantage of this technology is the complete elimination of conventional formwork, therefore minimizing the amount of work required for the formwork assembly, as well as the amount of construction waste caused by the disposal of the used formwork [1]. Current

Materials 2020, 13, 1093; doi:10.3390/ma13051093 www.mdpi.com/journal/materials Materials 2020, 13, 1093 2 of 17 approaches of additive manufacturing in construction (AMC) are mainly based on the following methods: particle bed printing [2] and material extrusion [3]. A third emerging technology, the so-called Shotcrete 3D Printing, is based on material jetting [4]. Particle bed printing is a process where fine particles are selectively bound through the localized application of binder. For this, a thin layer of particles is evenly distributed inside a given building envelope. In a subsequent step a binder is deposited onto the particle layer, which selectively bonds the particles. Next, the particle bed is lowered, a new layer of particles is applied, and the binding process is repeated. In a layered manner, complex geometries can be fabricated, resting loosely inside the bed of unbound particles. This technology is used, for example, for the production of complex molds in metal foundry [5], but also in architecture as formwork for casting high resolution concrete parts [6]. Instead of using the printed element merely as formwork, there are also attempts to entirely print structural elements using large, room-sized particle bed printers [7]. In the extrusion-based approach to additive manufacturing, strands of fine concrete are printed layer-wise on top of each other, progressively creating a three-dimensional object. Here, predominantly two different concepts are pursued, firstly the contour crafting approach [8], and secondly the technique of 3D concrete printing [9]. Contour crafting is based on a lost formwork method where only the perimeter is printed, to create a hollow core element that is later filled with conventional concrete. Several research groups and industry professionals [10–13] have today adopted this method, also for the reason that it can be combined with the manual integration of reinforcement [14]. In the 3D concrete printing process, the entire concrete component is printed; however, channels for the placement of post-tensioning cables can be integrated in order to reinforce the component [15]. The third approach, Shotcrete 3D Printing, differs from extrusion-based concrete printing in the respect that the concrete is not deposited in strands, but is sprayed by means of compressed air [16]. This results in good layer adhesion, and the capability to spray around and hence embed structural reinforcement [17]. Each of these techniques features a distinct set of advantages, as well as process inherent limitations. Material extrusion, for example, offers high building rates. Depending on the printing system, state-of the-art concrete printers reach building speeds of up to 16 cm/s [18], corresponding to an extrusion volume flow of approximately 0.6 m3/h. Even higher volume flows of currently up to 1 m3/h can be achieved with the Shotcrete 3D Printing method. However, printing speed is inversely related to geometric resolution and surface quality. Accordingly, particle bed printing is, for example, the most geometrically versatile, but also the slowest additive manufacturing process. A common feature of all three approaches is that the printed object is always built-up in horizontal layers. In this paper a novel 3D printing technology with concrete is introduced, which challenges the layered build-up and proposes a printing approach enabling more complex spatial printing trajectories. In the so-called Injection 3D Concrete Printing (I3DCP) method, one fluid material (M1) is robotically injected into another fluid material (M2) with similar rheological properties. In consequence, the extruded material M1 maintains in a stable position within material M2 without sinking or buoying. Accordingly, this process combines the advantages of the two techniques described above: firstly, the material M1 can be extruded at high building rates, and secondly, the surrounding liquid enables omnidirectional printing and thus comprehensive spatial freedom without restriction by gravity.

2. Basic Principles of Injection 3D Concrete Printing In this paper the systematic investigation of three variations of Injection 3D Concrete Printing technique are described:

1. 3D printing of a ﬁne grain concrete into a non-hardening suspension called “CiS” (Concrete in Suspension); 2. 3D printing a non-hardening suspension into a ﬁne grain concrete called “SiC” (Suspension in Concrete); Materials 20192020, 1213, x 1093 FOR PEER REVIEW 3 of 17 Materials 2019, 12, x FOR PEER REVIEW 3 of 17 3. 3D Printing a fine grain concrete with specific properties into another fine grain concrete with

3.3. different3D3D PrintingPrinting properties aa finefine grain , grainthe so concrete concrete called “ with CwithiC” specific processspecific properties (Concreteproperties intoin into Concrete) another another. fine fine grain grain concrete concrete with with Withdidifferentfferent the properties, firstproperties process, the the v so ersionso called called, C “CiC” “iS,C iC a ” process fine process grain (Concrete (Concrete concrete in in Concrete).is Concrete) printed . into a non-hardening suspension.With theAs such, first processintricate vtrussersion structures, CiS, a fineand filigreegrain concrete concrete is space printed frames into can a nonbe fabricated-hardening, With the first process version, CiS, a fine grain concrete is printed into a non-hardening suspension. whichsuspension. are otherwise As such, difficult intricate or impossibletruss structures to manufacture and filigree. Figure concrete 1 shows space the frames schematic can be fabrication fabricated , As such, intricate truss structures and filigree concrete space frames can be fabricated, which are methodwhich ,are as wellotherwise as a possibly difficult printed or impossible object. to manufacture. Figure 1 shows the schematic fabrication otherwisemethod, as di ffiwellcult as or a impossiblepossibly printed to manufacture. object. Figure1 shows the schematic fabrication method, as well as a possibly printed object.

(a) (b) (a) (b) Figure 1. Concrete in Suspension (CiS): (a) fabrication process diagram; (b) printed lattice structure. FigureFigure 1. 1.Concrete Concrete in in Suspension Suspension (CiS): (CiS): ( a()a) fabrication fabrication process process diagram; diagram; ( b(b)) printed printed lattice lattice structure. structure. The second process version SiC (suspension in concrete) is the reversal of the CiS principle: in the SiCTheThe process second second, process aprocess non-hardening versionversion SiCS isuspensionC (suspension is printedin concrete) concrete) into is ais the ve thessel reversal reversal filled of withof the the CiS fresh CiS principle principle: concrete: i.n Subsequentlyinthe the SiC SiC process process,, the, a suspension a non non-hardening-hardening is removed, suspension suspension leaving is is printed printed cavities into into or a channels. vesselvessel ﬁlledfilled This with with can fresh fresh be concrete. used concrete to . functionallySubsequently,Subsequently grad the, thee suspension concrete suspension components is removed, is removed, through leaving leaving cavitiesdifferentiated cavities or channels. density or channels. This or to can utilize beThis used them can to forfunctionally be physical used to elementgradefunctionally concrete activation grad componentse. concreteFigure 2 components throughshows schematically diﬀ erentiatedthrough differentiated the density SiC-fabrication or to density utilize proce or them toss utilize for as well physical them as for a element printedphysical element.activation.element activation Figure2 shows. Figure schematically 2 shows schematically the SiC-fabrication the SiC process-fabrication as well proce as ass printed as well element. as a printed element.

(a) (b) (a) (b) FigureFigure 2. SuspensionSuspension in in Concrete Concrete (SiC): (SiC): ( (aa)) f fabricationabrication process diagram; ( b) printed graded element element.. Figure 2. Suspension in Concrete (SiC): (a) fabrication process diagram; (b) printed graded element. For the third version of this concept concept,, concrete is used for both the extrusion and the supporting For the third version of this concept, concrete is used for both the extrusion and the supporting material, whereas whereas the the material material properties properties differ diﬀ eraccording according to functional to functional needs needs (CiC (CiC,, Figure Figure 3). One3). material, whereas the material properties differ according to functional needs (CiC, Figure 3). One possibilityOne possibility is to isinject to inject a high a- high-performanceperformance concrete concrete with with high high strength strength into ainto concrete a concrete with with low possibility is to inject a high-performance concrete with high strength into a concrete with low strength.low strength. When When injecting injecting along alongthe prevailing the prevailing compression compression trajectories trajectories,, a severely a severely improved improved overall strength. When injecting along the prevailing compression trajectories, a severely improved overall mechanicaloverall mechanical performance performance of the printed of the printedelement elementis achievable is achievable.. The special The feature special of feature the CiC of method the CiC mechanical performance of the printed element is achievable. The special feature of the CiC method comparedmethod compared to the other to the variants other variantsis that both is that liquids both form liquids a permanent form a permanent compound. compound. compared to the other variants is that both liquids form a permanent compound.

Materials 2020, 13, 1093 4 of 17 Materials 2019, 12, x FOR PEER REVIEW 4 of 17

(a) (b)

Figure 3.3. Concrete in Concrete (CiC): (a) fabricationfabrication process diagram; (b) sectionsection throughthrough aa locallylocally strengthenedstrengthened element.element.

3. State of the Art: The underlying concept,concept, to continuously extrudeextrude material in spatial trajectories,trajectories, was previously investigatedinvestigated with other materials,materials, withwith andand withoutwithout aa supportingsupporting medium.medium. TheThe MataerialMataerial researchresearch projectproject of the Institute for Advanced Advanced Architecture Architecture of of Catalonia Catalonia (IAAC (IAAC)) investigated investigated the the us usageage of of a atwo two-component-component thermosetting thermosetting polymer polymer inin order order to to robotically robotically print print material material in in mid mid-air-air [19] [19].. Concurrently, the the Mesh Mesh Mould Mould research research at ETH at ZurichETH Zurich investigated investigated the use ofthe thermoplastic use of thermoplastic materials, formaterials, example, for Acrylnitril-Butadien-Styrol-Copolymereexample, Acrylnitril-Butadien-Styrol-Copolymere (ABS) and (ABS Polyactic) and P Acidolyactic (PLA), Acid to ( spatiallyPLA), to printspatially geometrically print geometrically complex meshcomplex structures mesh structures [20]. In both [20]. techniques, In both techniques, the ability the to ability print freelyto print in airfreely was in limited air was by limited the hardening by the hardening behavior behavior of the material of the material and the and gravitational the gravitational forces acting forces onacting the unsupportedon the unsupported material material cantilevering cantileve in thering air. in In the 2014, air. aIn research 2014, a team research from team Princeton from University Princeton demonstratedUniversity demonstrated a technique a called technique “buoyant called extrusion”, “buoyant inextrusion which a” thermoset, in which a material thermoset such material as urethane such plasticas urethane is printed plastic in ais container printed in with a container sodium carbomer with sodium gel [ 21carbomer]. The materials gel [21] can. The be materials adjusted socan that be theadjusted printed so materialthat the slowlyprinted rises, material sinks, slowly or floats rises, weightlessly sinks, or floats in the weightlessly other material. in the In theother Large-Scale material. RapidIn theLiquid Large Printing-Scale Rapid project, Liquid researchers Printing at the project, Massachusetts researchers Institute at the of Technology Massachusetts are investigating Institute of theTechnology potential are of investigating this method forthe thepotential large-scale of this industrial method for production the large-scale of objects industrial using production high-quality of industrial-gradeobjects using high rubbers,-quality foams, industrial and- plasticsgrade rubbers, [22]. foams, and plastics [22]. These initial initial successes successes using polymers polymers triggered testing of of the the concept concept with with cement-bonded cement-bonded materials asas well.well. ForFor example,example, thethe FrenchFrench start-upstart-up Soliquid Soliquid recently recently published published a a video video of of extruding extruding a concretea concrete in in a containera container filled filled with with gel gel [23 [23]]. Another. Another team team of of researchers researchers from fro Singaporem Singapore University University of Technologyof Technology and andDesign Design (SUTD) (SUTD) robotically robotically injected injected a reactive a reactive gas-forming gas-forming powder, intopowder, a flat intoformwork a flat filledformwork with filled concrete, with in concrete, order to in create order porous to create building porous elements building [24 elements]. These [24] experiments. These experiments concurred withconcurred the initial with experiments the initial experiments at TU Braunschweig at TU Braunschweig and the filing and of the the filing multi-material of the multi patent-material “Injection patent 3D“Injection Printing”, 3D relatedPrinting” to, thisrelated paper to [this25]. paper [25] . From the outset, the concrete-basedconcrete-based I3DCPI3DCP research at TU Braunschweig aimed to systematically investigateinvestigate the method inin itsits threethree conceptualconceptual variations—CiS,variations—CiS, SiC,SiC, and CiC—inCiC—in order to create a coherent systemsystem andand toto develop material combinations thatthat make all three versions applicable for the construction industry.industry. Here,Here, aa particularparticular focusfocus waswas directeddirected to:to:  the identification of process inherent applications in construction; the identification of process inherent applications in construction; •  the use and design of materials suitable for construction in terms of robustness, ecological, the use and design of materials suitable for construction in terms of robustness, ecological, and • and economic factors; economic fundamental factors; interactions between material behavior and fabrication process. fundamental interactions between material behavior and fabrication process. • 4. Experimental Setup In the following sections, the fundamental and systematic exploration of the three versions of I3DCP is described. The methods were developed in the research seminar “Digital Building Fabrication Studio” (DBFS) at TU Braunschweig and were validated through a manifold of physical

Materials 2020, 13, 1093 5 of 17

4. Experimental Setup

MaterialsIn the 2019 following, 12, x FOR PEER sections, REVIEW the fundamental and systematic exploration of the three versions5 of 17 of I3DCP is described. The methods were developed in the research seminar “Digital Building Fabrication Studio”experiments. (DBFS) T athe TUfollowing Braunschweig sections anddescribe were the validated robotic throughsetup used a manifold, the digital of workflows physical experiments., and the Thematerials following developed sections. This describe is followed the robotic by a setup description used, of the the digital experiments workflows, on each and of the the materials three variations of the I3DCP technique. developed. This is followed by a description of the experiments on each of the three variations of the I3DCP technique. 4.1. Robotic Setup 4.1. RoboticAll experiments Setup were carried out on an UR 10 robot [26] using one of two custom-designed extruders. One of the extruders was based on a pneumatic material feed system, the other on a All experiments were carried out on an UR 10 robot [26] using one of two custom-designed material feed system actuated by an electromechanically driven piston. Both end effectors featured extruders. One of the extruders was based on a pneumatic material feed system, the other on a material specific advantages as well as limitations. The pneumatic extruder was robust and easy to handle, feed system actuated by an electromechanically driven piston. Both end effectors featured specific consisting of only a few components. The disadvantage, however, was that the material was fed advantages as well as limitations. The pneumatic extruder was robust and easy to handle, consisting indirectly by air pressure, and was therefore difficult to regulate precisely. In contrast, the ofelectromechanic only a few components. extruder Theallow disadvantage,ed for defined however, control ofwas the that extrusion the material volume was by fed positioning indirectly the by air pressure,piston with and wassub-millimeter therefore di precision.fficult to regulateAs a result, precisely. the extrusion In contrast, process the electromechanic could be stopped extruder and allowedrestarted for at defined any time. control This of theability extrusion was additionally volume by positioning enhanced by the the piston implementation with sub-millimeter of a precision.pneumatically As a result,actuated the pinch extrusion valve process, mounted could between be stopped the concrete and restarted hose and at the any nozzle time.. Both This end ability waseffectors additionally had a material enhanced capacity by the of implementation 8 liters. Due to ofrobot a pneumatically payload limitations, actuated the pinch material valve, cartridge mounted betweenwas decoupled the concrete from hose the robot and theand nozzle.was stationary Both end-mounted. effectors An had elongated a material interchangeable capacity of steel 8 liters. Duenozzle to robotwith a payload length of limitations, 80 cm and varying the material diameters cartridge from 8 to was 16 mm decoupled was connected from the to the robot cartridge and was stationary-mounted.by a flexible rubber hose, An elongated and was attached interchangeable to the robot steel via nozzle a 3D withprinted a length adapter of. 80Both cm the and electric varying diametersand the pneu frommatic 8 to 16extruder mm was were connected controlled to thevia cartridgethe analogue by a and flexible digital rubber outputs hose, of andthe robot. was attached The torobotic the robot setup via is a depicted 3D printed in Figure adapter. 4. Both the electric and the pneumatic extruder were controlled via the analogue and digital outputs of the robot. The robotic setup is depicted in Figure4.

Figure 4. Robotic setup: electromechanical and pneumatic extruder, hose, nozzles with different Figure 4. Robotic setup: electromechanical and pneumatic extruder, hose, nozzles with diﬀerent diameters, and robot with nozzle. diameters, and robot with nozzle.

4.2.4.2. Design Design and and Control Control ForFor design, design, path path planning, planning, simulation, simulation, and and robot robot control control an an integrative integrative design design-to-fabrication-to-fabrication approachapproach was was developed. developed. ForFor thisthis purpose, the the modeling modeling software software Rhino Rhino 3D 3D [27] [27 with] with its itsintegrated integrated programmable,programmable, graphical graphical extensionextension GrasshopperGrasshopper was was used. used. Inside Inside the the Grasshopper Grasshopper environment, environment, purpose-built plug-ins can be implemented or programmed for specific applications. Here, the purpose-built plug-ins can be implemented or programmed for speciﬁc applications. Here, the “Robots” “Robots” plugin [28], developed at the Bartlett School of Architecture, which contains a library of plugin [28], developed at the Bartlett School of Architecture, which contains a library of widely used widely used robotic set-ups, was used to simulate and control the robot. Within this software robotic set-ups, was used to simulate and control the robot. Within this software environment, environment, a continuous design-to-fabrication workflow was implemented in which the design a continuous design-to-fabrication workﬂow was implemented in which the design geometry is geometry is automatically converted into a simulation and subsequently into robot instructions. automatically converted into a simulation and subsequently into robot instructions. 4.3. Materials There were three materials used in the present experiments: a) fine grain concrete with an Ordinary Portland Cement (OPC, CEM I 52.5N) and a quartz sand with a maximum grain size of 2

Materials 2020, 13, 1093 6 of 17

4.3. Materials There were three materials used in the present experiments: (a) ﬁne grain concrete with an Ordinary Portland Cement (OPC, CEM I 52.5N) and a quartz sand with a maximum grain size of 2 mm; (b) ground limestone-sand suspension with a maximum grain size of 0.5 mm as non-hardening suspension; and (c) sand-gel suspension as non-hardening suspension with a maximum grain size of 0.5 mm. A detailed overview of all components used is given in Table1, below. Table2 summarizes the fabrication setup, parameters and materials used in the diﬀerent experimental investigations.

Table 1. Mixture compositions of used material.

Limestone-Sand Fine Grain Concrete Sand-Gel Suspension Suspension Components (kg/m3) (kg/m3) (kg/m3) Portland Cement (CEM I 52.5N) 450.8 - - Ground limestone (LS) - 695.6 - Puzzolan 144.3 - - Silica fume 27.1 - - Aggregate, d = 0–0.5 mm - 1237.0 727.1 Aggregate, d = 0–2 mm 1064.0 - - Water 342.8 275.6 - Ultra Sonic gel - - 727.1 PP-Fibres 2.7 - - Additives (solid) 29.8 - - Superplasticizer (liquid) - 1.0 - Pigment (Iron Oxide, Black) 1 10.08 - - 1: Used only for the extruded material in the CiC process

Table 2. Overview of the fabrication setup, parameters and materials used during the diﬀerent investigations.

1 Preliminary Robot Extrusion Injection Studies 2 Extruder Support Material Speed Volume Material Demonstrator 1 Gel and sand-gel Fine grain suspension CiS pneumatic 5 m/min 1 l/min (0.5 bar) concrete 2 Limestone- sand-suspension Gel-sand SiC electro-mechanic 5 m/min 1 l/min Fine grain concrete suspension Pigmented ﬁne CiC pneumatic 5 m/min 1 l/min (0.5 bar) Fine grain concrete grain concrete 1 Material used during the exploratory phase of the process. 2 Material used for the fabrication of the ﬁnal demonstrators.

5. Investigations, Results, and Discussion

5.1. Concrete in Suspension (CiS) Printing concrete in a suspension offers the potential to create lightweight, intricate spaceframe structures, with a high strength to weight ratio, which are difficult to manufacture in other ways. The focus of these investigations was directed firstly toward the design potentials of this approach, Materials 2019, 12, x FOR PEER REVIEW 7 of 17

MaterialsPrinting2020, 13 concrete, 1093 in a suspension offers the potential to create lightweight, intricate spaceframe7 of 17 structures, with a high strength to weight ratio, which are difficult to manufacture in other ways. The focus of these investigations was directed firstly toward the design potentials of this approach, and secondlyand secondly to develop to develop material material formulations formulations that thatare technically are technically and andeconomically economically feasible feasible,, also alsofor the for harshthe harsh environment environment of the of construction the construction site. site.

5.1.1. CiS CalibrationCalibration ProcessProcess and PreliminaryPreliminary StudiesStudies For calibration purposes purposes a a series series of of small small-scale-scale experiments experiments was was conducted. conducted. In In order order to toallow allow a visuala visual examination examination of of the the extrusion extrusion process process,, the the fine fine grain grain concrete concrete was was printed printed into into a a transparent vessel (37.6 26 18.9 cm) filled with gel. The pneumatically actuated end effector was filled with vessel (37.6 × 26 ×× 18.9 cm) filled with gel. The pneumatically actuated end effector was filled with 3 lit3 litersers of of fine fine grain grain concrete concrete,, and the airair pressurepressure was was initially initially set set to to 2 bar.2 bar A. simple,A simple, yet yet spatial, spatial zig, zagzig zaggeometry geometry was was chosen chosen for thefor initialthe initial experiment. experiment. The printed structure was removed from the gel after 48 hours of curing.curing. Already during th thisis time,time, the following observations were made made:: only only a a few few hours hours after after printing printing,, a a white, white, dense dense,, gel gel-skin-skin developed around the concrete structure (Figure 55)).. AtAt thethe samesame time,time, waterwater accumulatedaccumulated onon thethe surfasurfacece of the gel. In In this this respect, respect, it it can can be be expected expected that that the the absolute absolute position position of of the the object object in in the the gel gel changed, and thatthat thethe gelgel deteriorateddeteriorated while while in in contact contact with with the the curing curing concrete. concrete. The The decomposition decomposition of ofthe the gel gel restricts restricts the the reusability reusability of theof the material, material, and and approximately approximately 5% losses5% losses can becan expected be expected after after each eachprinting printing process. process.

(a) (b)

Figure 5. Initial experimentexperiment forfor thethe purpose purpose of of calibration: calibration: (a ()a zig) zig zag zag geometry geometry printed printed into into gel, gel, (b) ( theb) theobject object after after 48hours 48 hours of curing of curing inside inside the the gel. gel.

As the extrusion extrusion appeared very bulky and undefined, undefined, i inn the subsequent subsequent calibration steps steps,, the air air pressure waswas graduallygradually reduced,reduced, while while the the robot robot speed speed was was kept kept constant. constant. This This was was repeated repeated until until the therobot robot speed speed and and extrusion extrusion rate rate resulted resulted in a in constant a constant extrusion extrusion diameter diameter corresponding corresponding to the to 16the mm 16 mmdiameter diameter of the of nozzle. the nozzle. For this,For this, an air an pressure air pressure of 0.5 of bar 0.5 was bar found was found to be ato suitable. be a suitable. Different Different space spacefilling filling geometries geometries were testedwere teste (Figured (Figure6). 6).

(a) (b)

Materials 2019, 12, x FOR PEER REVIEW 7 of 17

Printing concrete in a suspension offers the potential to create lightweight, intricate spaceframe structures, with a high strength to weight ratio, which are difficult to manufacture in other ways. The focus of these investigations was directed firstly toward the design potentials of this approach, and secondly to develop material formulations that are technically and economically feasible, also for the harsh environment of the construction site.

5.1.1. CiS Calibration Process and Preliminary Studies For calibration purposes a series of small-scale experiments was conducted. In order to allow a visual examination of the extrusion process, the fine grain concrete was printed into a transparent vessel (37.6 × 26 × 18.9 cm) filled with gel. The pneumatically actuated end effector was filled with 3 liters of fine grain concrete, and the air pressure was initially set to 2 bar. A simple, yet spatial, zig zag geometry was chosen for the initial experiment. The printed structure was removed from the gel after 48 hours of curing. Already during this time, the following observations were made: only a few hours after printing, a white, dense, gel-skin developed around the concrete structure (Figure 5). At the same time, water accumulated on the surface of the gel. In this respect, it can be expected that the absolute position of the object in the gel changed, and that the gel deteriorated while in contact with the curing concrete. The decomposition of the gel restricts the reusability of the material, and approximately 5% losses can be expected after each printing process.

(a) (b)

Figure 5. Initial experiment for the purpose of calibration: (a) zig zag geometry printed into gel, (b) the object after 48 hours of curing inside the gel.

As the extrusion appeared very bulky and undefined, in the subsequent calibration steps, the air pressure was gradually reduced, while the robot speed was kept constant. This was repeated until the robot speed and extrusion rate resulted in a constant extrusion diameter corresponding to the 16 Materialsmm diameter2020, 13, of 1093 the nozzle. For this, an air pressure of 0.5 bar was found to be a suitable. Different8 of 17 space filling geometries were tested (Figure 6).

Materials 2019, 12, x FOR PEER REVIEW 8 of 17 (a) (b)

Figure 6. Results of the calibration calibration p process:rocess: ( a)) a a spiraling spiraling geometry geometry printed in gel; ( b) a more more complex geometry printed in a 1:1 gel-sandgel-sand suspension.

In order to to prevent prevent the the decomposition decomposition of of the the gel, gel, as aswell well as the as theformation formation of the of white the white skin, skin, and andto counteract to counteract progressive progressive sinking sinking of the of printed the printed structur structure,e, two alternative two alternative support support mediums mediums were weredeveloped developed and tested and tested.. Firstly, Firstly, in an effort in an to e ﬀachieveort to achieve greater greateroverall stability overallstability (in terms (in of termsthe material of the materialdecomposition decomposition as well as as printing well as printingsupport) support),, the same the bulk same volume bulk volume of quartz of quartzsand was sand added was addedto the topure the gel. pure Ad gel.ditionally Additionally,, a less a less sensitive sensitive and and more more cost cost-e-effectiveﬀective limestone limestone-sand-sand suspension suspension was developed, which diddid not require thethe useuse ofof expensiveexpensive gelgel atat all.all. Both mediums were tested and compared in printingprinting experimentsexperiments (Figure(Figure7 7).).

(a) (b)

Figure 7. ObjectsObjects printed printed in in two two different diﬀerent support support mediums: mediums: (a)( printeda) printed in a in gel a- gel-sandsand suspension; suspension; (b) (printedb) printed in a in limestone a limestone powder powder quartz quartz sand sand suspension suspension..

For the gel-sandgel-sand suspension no formation of a white skin was observed, and only a very small amount of water accumulated on the surfacesurface of thethe printingprinting container.container. When extruding within the limestone-sandlimestone-sand suspension suspension,, it it was observed that that furrows were created behind the printing nozzle, when moving moving through through the the gel. gel. Those Those furrows furrows saturated saturated with with concrete, concrete, leading leadi tong the to creation the creation of ridges of aboveridges theabove print the trajectory print trajectory (Figure (Figure7b). In 7 furtherb). In further experiments experiments this was this correctedwas corrected by increasing by increasing the waterthe water content content by 3%. by In3% contrast. In contrast to the to gel-based the gel- suspensions,based suspensions, no problems no problems with water with going water through going thethrough limestone-sand the limestone suspension-sand suspension on the suspension’s on the suspension surface’s were surface observed were observed at all. at all.

5.1.2. CiS D Demonstratoremonstrator To validate the CiS principle, a medium-sized object was developed for production in a 50 50 To validate the CiS principle, a medium-sized object was developed for production in a 50× × 50× 50× 50 cm cm container. container. For For this, this a, a formwork formwork was was ﬁlled filled with withapproximately approximately 110110 litersliters ofof thethe non-hardeningnon-hardening limestone-sandlimestone-sand suspension.suspension. The The object object represents represents a 14-sided a 14-sided polyhedral polyhedral spaceframe spaceframe structure structure derived fromderived a hexagonal from a hexagonal base (Figure base8 a).(Figure 8a).

(a) (b) (c)

Materials 2019, 12, x FOR PEER REVIEW 8 of 17

Figure 6. Results of the calibration process: (a) a spiraling geometry printed in gel; (b) a more complex geometry printed in a 1:1 gel-sand suspension.

In order to prevent the decomposition of the gel, as well as the formation of the white skin, and to counteract progressive sinking of the printed structure, two alternative support mediums were developed and tested. Firstly, in an effort to achieve greater overall stability (in terms of the material decomposition as well as printing support), the same bulk volume of quartz sand was added to the pure gel. Additionally, a less sensitive and more cost-effective limestone-sand suspension was developed, which did not require the use of expensive gel at all. Both mediums were tested and compared in printing experiments (Figure 7).

(a) (b)

Figure 7. Objects printed in two different support mediums: (a) printed in a gel-sand suspension; (b) printed in a limestone powder quartz sand suspension.

For the gel-sand suspension no formation of a white skin was observed, and only a very small amount of water accumulated on the surface of the printing container. When extruding within the limestone-sand suspension, it was observed that furrows were created behind the printing nozzle, when moving through the gel. Those furrows saturated with concrete, leading to the creation of ridges above the print trajectory (Figure 7b). In further experiments this was corrected by increasing the water content by 3%. In contrast to the gel-based suspensions, no problems with water going through the limestone-sand suspension on the suspension’s surface were observed at all.

5.1.2. CiS Demonstrator To validate the CiS principle, a medium-sized object was developed for production in a 50 × 50 × 50 cm container. For this, a formwork was filled with approximately 110 liters of the non-hardening limestone-sand suspension. The object represents a 14-sided polyhedral spaceframe structure Materials 2020, 13, 1093 9 of 17 derived from a hexagonal base (Figure 8a).

Materials 2019(a) , 12, x FOR PEER REVIEW (b) (c) 9 of 17

FigureFigure 8. Object 8. Object for process for process veriﬁcation: verification: (a) rendered (a) rendered structure, structure, (b) wireframe (b) wireframe model model of the of printing the printing path,path, (c) robot (c) robot simulation. simulation.

The printing path and sequence, as well as the robot simulation, are depicted in Figure8b,c. The printing path and sequence, as well as the robot simulation, are depicted in Figure 8b and The total path length was calculated to be 13.4 meters, resulting in an estimated printing time of c. The total path length was calculated to be 13.4 meters, resulting in an estimated printing time of 2:41 minutes, printed with a robot speed of 5 m/min. The printing process is schematically shown in 2:41 minutes, printed with a robot speed of 5 m/min. The printing process is schematically shown in Figure9. Figure 9.

(a) (b)

FigureFigure 9. Printing 9. Printing process process in the in limestone-sandthe limestone-sand suspension: suspension (a): closeup; (a) closeup; (b) general (b) general setup setup with with formworkformwork placed placed on the on ground. the ground.

5.1.3.5.1.3. CiS ResultsCiS Results AfterAfter three three days days the the object object was was removed removed from from the the limestone-sand limestone-sand suspensionsuspension ( (FigureFigure 1010a).a). The Thevast majority of node nodess w wereere connected monolithically (24 (24 out out of of 27), 27), and and the the overall overall structure structure was wasstable. stable. A A few few node nodess however, however, especially especially on on one one side side of ofthe the upper upper edge, edge, did didnot notconnect connect well well,, leaving leavingsmall small gap gapss between between the thespaceframe spaceframe elements elements.. One One node node was was entirely entirely disconnected disconnected from from the the edge edgeabove above,, suggesting suggesting that that the the position position of ofthe the material material changed changed or was or was actively actively displaced displaced during during printing printing(Figure (Figure 10c). 10 Inc). addition, In addition, a slight a slight expression expression of of ridges ridges above above the the extrusion extrusion was was observed observed in in this this experiment ( (FigureFigure 10bb).). This This indicates that that also in this experiment the supporting suspension was wastoo stiff stiff, and concrete flowedflowed intointo thethe furrowsfurrows thatthat thethe nozzlenozzle hadhad createdcreated whilewhile moving.moving. Hence,Hence, the the yieldyield stressstress andand viscosityviscosity of the suspension needs needs to to be be decreased decreased,, while while a a hig higherher viscosity viscosity and and yield yieldstress stress of of the the injected injected fine fine grain grain mortar mortar would would reduce reduce the the appearance appearance of of the the ridges. ridges.

(a) (b) (c)

Figure 10. CiS demonstrator: (a) stripping the object from the formwork; (b) final object; (c) detailed view of a knot with a deficient connection.

5.2. Investigations on Suspension in Concrete (SiC) Materials 2019, 12, x FOR PEER REVIEW 9 of 17

Figure 8. Object for process verification: (a) rendered structure, (b) wireframe model of the printing path, (c) robot simulation.

The printing path and sequence, as well as the robot simulation, are depicted in Figure 8b and c. The total path length was calculated to be 13.4 meters, resulting in an estimated printing time of 2:41 minutes, printed with a robot speed of 5 m/min. The printing process is schematically shown in Figure 9.

(a) (b)

Figure 9. Printing process in the limestone-sand suspension: (a) closeup; (b) general setup with formwork placed on the ground.

5.1.3. CiS Results After three days the object was removed from the limestone-sand suspension (Figure 10a). The vast majority of nodes were connected monolithically (24 out of 27), and the overall structure was stable. A few nodes however, especially on one side of the upper edge, did not connect well, leaving small gaps between the spaceframe elements. One node was entirely disconnected from the edge above, suggesting that the position of the material changed or was actively displaced during printing (Figure 10c). In addition, a slight expression of ridges above the extrusion was observed in this experiment (Figure 10b). This indicates that also in this experiment the supporting suspension was Materialstoo stiff,2020 and, 13 concrete, 1093 flowed into the furrows that the nozzle had created while moving. Hence, the10 of 17 yield stress and viscosity of the suspension needs to be decreased, while a higher viscosity and yield

(a) (b) (c)

Figure 10. 10. CiSCiS demonstrator: demonstrator: (a) ( astripping) stripping the the object object from from the theformwork; formwork; (b) final (b) ﬁnal object; object; (c) detailed (c) detailed view of of a a knot knot with with a adeficient deﬁcient connection. connection.

5.2. Investigations Investigations on on Suspension Suspension in in Concrete Concrete (Si (SiC)C) The potential of the SiC approach is to create concrete elements which are graded in porosity, or which can integrate geometrically complex systems of channels or ducts. These features could, for example, be used for aspects of structural performance, material savings, technical installations, or for component activation and building physics. In this approach the voids are created by printing a non-hardening suspension acting as displacement bodies directly into the fresh concrete. The challenges here were firstly, calibrating the material properties in such a way that the material remained stable within the concrete and generated well-defined voids, and secondly, designing the channel layout in a manner that the displacement material could drain off after the concrete had cured.

5.2.1. SiC Calibration Process and Preliminary Studies A preliminary setup was developed for material and process calibration that allowed immediate visual feedback. For this purpose, a 30 30 4 cm3 formwork was built, in which the 30 30 cm2 × × × sides were closed with Plexiglas. For an initial test a centrically placed regular 6 6-point grid was × generated, and at each point gel was injected into the fine grain concrete for a certain amount of time. The electromechanic piston extruder was used, due to the advantage that the material volume can be controlled more precisely. Compared to the pneumatic extruder, the material flow could be stopped and started at any time, whereas the pneumatic extruder continually builds up or releases pressure. This is particularly important if only material is to be injected selectively, for example as discretely enclosed volumes or voids. However, the initial experiment showed that simply stopping the piston does not automatically stop the material from flowing. Therefore, as a side-effect, channels were created when moving the nozzle from one point to the next. Moreover, it was recognized that the gel did not have the capacity (density) to displace the concrete evenly (Figure 11b) and that the same white skin as in the CiS approach appeared inside the voids (Figure 11c). As a consequence of these findings, firstly the pneumatic pinch valve for abruptly stopping the material flow was implemented, and secondly the alternative mediums (sand-gel suspension and limestone-sand suspension) were subsequently tested. During these tests it was observed that the limestone-sand suspension was not easily extrudable and segregated when pressure was exerted. Hence, for the following experiments a gel-sand mix was tested and a ratio of 1:0.96 was found to be ideal. Materials 2019, 12, x FOR PEER REVIEW 10 of 17

The potential of the SiC approach is to create concrete elements which are graded in porosity, or which can integrate geometrically complex systems of channels or ducts. These features could, for example, be used for aspects of structural performance, material savings, technical installations, or for component activation and building physics. In this approach the voids are created by printing a non-hardening suspension acting as displacement bodies directly into the fresh concrete. The challenges here were firstly, calibrating the material properties in such a way that the material remained stable within the concrete and generated well-defined voids, and secondly, designing the channel layout in a manner that the displacement material could drain off after the concrete had cured.

5.2.1. SiC Calibration Process and Preliminary Studies A preliminary setup was developed for material and process calibration that allowed immediate visual feedback. For this purpose, a 30 × 30 × 4 cm³ formwork was built, in which the 30 × 30 cm² sides were closed with Plexiglas. For an initial test a centrically placed regular 6 × 6-point grid was generated, and at each point gel was injected into the fine grain concrete for a certain amount of time. The electromechanic piston extruder was used, due to the advantage that the material volume can be controlled more precisely. Compared to the pneumatic extruder, the material flow could be stopped and started at any time, whereas the pneumatic extruder continually builds up or releases pressure. This is particularly important if only material is to be injected selectively, for example as discretely enclosed volumes or voids. However, the initial experiment showed that simply stopping the piston does not automatically stop the material from flowing. Therefore, as a side-effect, channels were created when moving the nozzle from one point to the next. Moreover, it was recognized that the gel did not have the capacity (density) to displace the concrete evenly (Figure 11b) and that the same white skin as in the CiS approach appeared inside the voids (Figure 11c). As a consequence of these findings, firstly the pneumatic pinch valve for abruptly stopping the material flow was implemented, and secondly the alternative mediums (sand-gel suspension and limestone-sand suspension) were subsequently tested. During these tests it was Materialsobserved2020 , that13, 1093 the limestone-sand suspension was not easily extrudable and segregated when11 of 17 pressure was exerted. Hence, for the following experiments a gel-sand mix was tested and a ratio of 1:0.96 was found to be ideal.

(a) (b) (c)

FigureFigure 11. 11.Initial Initial experimentexperiment forfor calibration:calibration: ( (aa)) 6 6 × 66-point-point pattern; pattern; (b (b) )result result of of the the injection injection process; process; × (c()c closeup) closeup of of the the irregular irregular voidsvoids withwith whitewhite skinskin and the unintended channels channels above. above.

5.2.2.5.2.2. SiC SiC Demonstrator Demonstrator InIn order order to to demonstrate demonstrate thethe uniqueunique potentials of of the the SiC SiC process, process, a a series series of of demonstrators demonstrators were were developeddevelopedand and realized.realized. The The design design of of a perforated a perforated facade facade panel panel was wasselected selected as the as architectural the architectural case casestudy. study. For Forthis, this, a larger a larger format format of 50 of × 5050 cm502 was cm2 chosen,was chosen, whereas whereas the thickness the thickness of the ofpanel the panelwas × wasmaintained. maintained. In contrast In contrast to the to the preliminary preliminary tests, tests, in thisin this serie seriess a differentiation a differentiation of of the the size size of of the the displacementdisplacement bodies bodies waswas deliberatelydeliberately made and reali realizedzed by by varying varying the the duration duration of of the the extrusion extrusion at at Materials 2019, 12, x FOR PEER REVIEW 11 of 17 each point. Moreover, the pattern was changed from a regular 6 6 configuration to a diagonal pattern, × whereeach point one row. Moreover, of points thewas pattern shifted horizontally was changed by from half the a regular distance 6 between × 6 configuration two points (Figureto a diagonal 12a,b). Thepattern, point where pattern one was row computationally of points was generated,shifted horizontally and automatically by half sortedthe distance from bottom between to top,two sopoints that it(Figure could 12a be printedand b). The without point self-intersections. pattern was computationally For each of generated, the demonstration and automatically panels the sorted difference from betweenbottom to the top, total so that volume it cou ofld the be displacementprinted without bodies self-intersections. and the formwork For each capacity of the wasdemonstration calculated. Thepanels corresponding the difference quantity between of the concrete total volume was then of the prepared displacement andpoured bodies and into the the formwork formwork capacity before printingwas calculated. (Figure 12 Thec). corresponding quantity of concrete was then prepared and poured into the

(a) (b) (c)

FigureFigure 12. ObjectObject for for process process verification, veriﬁcation, SiC: SiC: (a ()a shifted) shifted pattern; pattern; (b ()b path) path panning panning and and simulation; simulation; (c) (fabricationc) fabrication process process injecting injecting gel-sand gel-sand into into a fine a ﬁne grain grain concrete. concrete.

5.2.3.5.2.3. SiC ResultsResults ByBy usingusing thethe gel-sandgel-sand mixturemixture instead of thethe purepure gelgel forfor displacingdisplacing thethe concrete,concrete, signiﬁcantlysignificantly betterbetter resultsresults werewere achieved.achieved. The The concrete concrete was was displaced evenly, evenly, creating almost spherical voids voids.. SlightSlight geometrical distortionsdistortions werewere possiblypossibly duedue toto thethe fact that the suspension was not printed freely intointo thethe concrete,concrete, butbut againstagainst thethe plexiglassplexiglass panel.panel. In orderorder toto verifyverify thisthis assumption,assumption, aa volumetricvolumetric sample would have to be produced and then cut open. In addition to the voids, some panels also had an integrated channel system that connected the voids. In this specific demonstrator those channels were intended to be used as an integrated watering system for façade greening elements of physical component activation (Figure 13b).

(a) (b) (c)

Figure 13. SiC demonstrators: (a) panel with displacement bodies gradually decreasing in size towards the top; (b) concept for façade planting; (c) integration of continuous channels connecting the voids for watering.

5.3. Investigations Concrete in Concrete (CiC) The potential of the CiC method is the ability to locally strengthen a concrete building element by injecting a high strength concrete into a lower grade material, for example, recycled concrete. In this process the high-strength concrete is injected along the compression trajectories within a concrete component. This local reinforcement makes it possible to construct more efficient structures, as the Materials 2019, 12, x FOR PEER REVIEW 11 of 17

each point. Moreover, the pattern was changed from a regular 6 × 6 configuration to a diagonal pattern, where one row of points was shifted horizontally by half the distance between two points (Figure 12a and b). The point pattern was computationally generated, and automatically sorted from bottom to top, so that it could be printed without self-intersections. For each of the demonstration panels the difference between the total volume of the displacement bodies and the formwork capacity was calculated. The corresponding quantity of concrete was then prepared and poured into the formwork before printing (Figure 12c).

(a) (b) (c)

Figure 12. Object for process verification, SiC: (a) shifted pattern; (b) path panning and simulation; (c) fabrication process injecting gel-sand into a fine grain concrete.

5.2.3. SiC Results

MaterialsBy2020 using, 13, 1093the gel-sand mixture instead of the pure gel for displacing the concrete, significantly12 of 17 better results were achieved. The concrete was displaced evenly, creating almost spherical voids. Slight geometrical distortions were possibly due to the fact that the suspension was not printed freely sampleinto the would concrete, have but to be against produced the plexiglass and then cut panel. open. In Inorder addition to verify to the this voids, assumption, some panels a volumetric also had ansample integrated would channel have to system be produced that connected and thenthe cut voids.open. In addition this speciﬁc to the demonstrator voids, some thosepanels channels also had were intended to be used as an integrated watering system for façade greening elements of physical component activation (Figure 13b).

(a) (b) (c)

FigureFigure 13. 13SiC. SiC demonstrators: demonstrators: (a) panel (a) panel with displacementwith displacement bodies bodies gradually gradually decreasing decreasing in size towards in size thetowards top; (b the) concept top; (b) for concept façade for planting; façade planting; (c) integration (c) integration of continuous of continuous channels channels connecting connecting the voids the forvoids watering. for watering.

5.3.5.3. Investigations Investigations Concrete Concrete in in Concrete Concrete (CiC) (CiC) TheThe potential potential of of the the CiC CiC method method is theis the ability ability to locallyto local strengthenly strengthen a concrete a concrete building building element element by injectingby injecting a high a high strength strength concrete concrete into ainto lower a lower grade grade material, material for example,, for example recycled, recycled concrete. concrete.In this In processthis process the high-strength the high-strength concrete concrete is injected is injected along along the the compression compression trajectories trajectories within within a a concrete concrete component.component. ThisThis locallocal reinforcement reinforcement makesmakes itit possible possible to to construct construct more more e efficientﬃcient structures, structures, as as the the internal structure can be topologically optimized and expensive high-strength material can be used, only where it delivers maximum eﬀect.

5.3.1. CiC Calibration Process and Preliminary Studies The aim of this experiment was to visualize the local diﬀerentiation of material properties, by printing a batch of pigmented concrete into a batch of the same, non-pigmented concrete. Here, 6 g of black pigment was added to every kg of dry premixed ﬁne grain concrete. Regarding the appropriate concrete rheological properties for both batches, it was possible to build on the experience already gained in the CiS and SiC process. Accordingly, for extruding the pigmented concrete, the pneumatic piston extruder was used.

5.3.2. CiC Demonstrator As proof of concept, a rectangular concrete beam measuring 80 15 20 cm3 was topologically × × optimized for a four-point bending scenario. Based on the resulting geometry, a continuous path was manually created (see Figure 14a). The total path length was measured to be 3.2 meters. With a robot speed of 5 m/min and a printing time of 39 seconds, a material consumption of 0.64 liters for printing was calculated. Accordingly, the formwork was filled with 23 liters of fine grain concrete, causing the formwork to be almost filled completely after the pigmented concrete was injected. Prior to pouring, two 10 mm reinforcement bars were placed in the tension zone of the beam, approximately 3 cm away from the outer surface. In order to avoid collisions with the reinforcement, the printing path was placed centrally within the formwork (Figure 14b). Materials 2019, 12, x FOR PEER REVIEW 12 of 17

internal structure can be topologically optimized and expensive high-strength material can be used, only where it delivers maximum effect.

5.3.1. CiC Calibration Process and Preliminary Studies The aim of this experiment was to visualize the local differentiation of material properties, by printing a batch of pigmented concrete into a batch of the same, non-pigmented concrete. Here, 6 g of black pigment was added to every kg of dry premixed fine grain concrete. Regarding the appropriate concrete rheological properties for both batches, it was possible to build on the experience already gained in the CiS and SiC process. Accordingly, for extruding the pigmented concrete, the pneumatic piston extruder was used.

5.3.2. CiC Demonstrator

As proof of concept, a rectangular concrete beam measuring 80 × 15 × 20 cm3 was topologically optimized for a four-point bending scenario. Based on the resulting geometry, a continuous path was manually created (see Figure 14a). The total path length was measured to be 3.2 meters. With a robot speed of 5 m/min and a printing time of 39 seconds, a material consumption of 0.64 liters for printing was calculated. Accordingly, the formwork was filled with 23 liters of fine grain concrete, causing the formwork to be almost filled completely after the pigmented concrete was injected. Prior to pouring, two 10 mm reinforcement bars were placed in the tension zone of the beam, approximately 3 cm Materials 2020, 13, 1093 13 of 17 away from the outer surface. In order to avoid collisions with the reinforcement, the printing path was placed centrally within the formwork (Figure 14b).

(a) (b)

FigureFigure 14. 14Object. Object for for process process veriﬁcation, verification, (CiC): (CiC): (a ()a topology) topology optimized optimized beam beam (top), (top), robotic robotic fabrication fabrication pathpath (bottom); (bottom); (b ()b simulation) simulation of of the the fabrication fabrication process. process.

5.3.3.5.3.3. CiC CiC Results Results TheThe printing printing process process forthis for demonstrator this demonstrator took merely took 39merely seconds. 39The seconds component. The was component demoulded was afterdemoulded 48 hours after of curing 48 hours and was of subsequently curing and was cut insubsequently transverse and cut longitudinal in transverse directions. and longitudinal The cuts showMaterialsdirections. that 2019 the, 12The printed, x FORcuts PEER geometryshow REVIEW that was the clearlyprinted deﬁned geometry and wa positioneds clearly correctlydefined and within positioned the cast elementcorrectly13 of 17 (Figurewithin 15 the). cast element (Figure 15).

FigureFigure 15. 15.Longitudinal Longitudinal sectionsection throughthrough thethe CiCCiC sample.sample.

6.6. PotentialsPotentials andand ChallengesChallenges ofof I3DCPI3DCP AllAll experimentsexperiments described described above above represent represent a a proof proof of of concept concept and and are are a a startingstarting pointpoint forfor furtherfurther researchresearch intointo InjectionInjection 3D3D ConcreteConcrete Printing.Printing. It It is is believed believed that that each each of of the the process process variations variations has has a aunique unique potential potential for for the the application application in in constructionconstruction.. This This is particularlyparticularly true as as these these initial initial tests tests showedshowed thatthat thethe processprocess isis fastfast andand thatthat thethe materialmaterial combinationscombinations developeddeveloped inin thisthis researchresearch areare inexpensiveinexpensive andand economicallyeconomically well within within the the range range of of common common building building materials materials.. For For example example,, by bysubstituting substituting the the pure pure gel gel sus suspensionpension (3.7 (3.7 €/kg)€/kg) with with a limestone a limestone-sand-sand suspension suspension (<0. (08<0.08 €/kg)€/,kg), the thecosts costs were were reduced reduced by byalmost almost two two orders orders of magnitude. of magnitude. Moreover, Moreover, all the all themethods methods presented presented in this in thispaper paper are are based based on on standard standard concrete concrete technologies, technologies, for for example example,, standard standard pumps and and formwork formwork systems.systems. Therefore, Therefore, all all processes processes are consideredare considered scalable scalable and, more and, importantly, more importantly, economically economically feasible onfeasible the large on the scale large of construction. scale of construction. In addition, In addition, the dimensional the dimensional limitations limitations are comparable are comparable to current to precastcurrent concrete precast concrete production production or in-situ or concrete in-situ constructions.concrete constructions. TheThe uniqueunique potentialspotentials ofof eacheach ofof thethe process process variation variation andand speciﬁcspecific applications applications inin constructionconstruction areare outlinedoutlined inin thethe following following paragraphs. paragraphs. TheThe challengeschallenges associatedassociated withwith up-scalingup-scaling mustmust alsoalso bebe identiﬁedidentified and and addressed addressed in in future future research. research.

6.1. CiS, Potentials and Challenges The CiS process offers the potential to radically expand the design space of concrete constructions. With contemporary techniques, be it conventionally casted or 3D printed, concrete elements are predominantly conceived as flat or curved surface structures (with the exception of columns). The CiS process, however, makes it possible to dissolve concrete structures into highly differentiated intricate spatial structures. This can be of practical use when, for example, structural performance is the main requirement, and other features, like, for example, the enclosure of a thermal envelope, are of secondary importance. In such cases the structure can be reduced to solely loadbearing members, and locally differentiated spaceframe structures can be printed with a minimum of material used. Moreover, and in addition to the potential of globally differentiating the spatial structures, the process enables the local differentiation of each individual truss member by altering the robot speed and extrusion rate during printing. First experiments have verified this approach, and are depicted in Figure 16a. Nevertheless, the structural performance of the systems depends not only on mere geometry but also on other parameters, like node connectivity, and on the integration of reinforcement. With the former, it must be ensured in each node that the materials bind monolithically to each other and that no separating layer is formed, as occasionally observed in the experiments. Regarding the integration of reinforcement, one major advantage here compared to the layered build-up of conventionally 3D printed structures, is the possibility to extrude material in line with the prevailing force trajectories. This unique feature makes it possible to co-extrude reinforcement, such as a continuous fiber or steel cables within the concrete strand. Initial tests have been carried out and are currently being developed further (Figure 16b).

Materials 2020, 13, 1093 14 of 17

6.1. CiS, Potentials and Challenges The CiS process offers the potential to radically expand the design space of concrete constructions. With contemporary techniques, be it conventionally casted or 3D printed, concrete elements are predominantly conceived as flat or curved surface structures (with the exception of columns). The CiS process, however, makes it possible to dissolve concrete structures into highly differentiated intricate spatial structures. This can be of practical use when, for example, structural performance is the main requirement, and other features, like, for example, the enclosure of a thermal envelope, are of secondary importance. In such cases the structure can be reduced to solely loadbearing members, and locally differentiated spaceframe structures can be printed with a minimum of material used. Moreover, and in addition to the potential of globally differentiating the spatial structures, the process enables the local differentiation of each individual truss member by altering the robot speed and extrusion rate during printing. First experiments have verified this approach, and are depicted in Figure 16a. Materials 2019, 12, x FOR PEER REVIEW 14 of 17

(a) (b)

FigureFigure 16. Further investigations: ( a) differentiated diﬀerentiated extrusion thickness; (b)) initial experiments testingtesting thethe co-extrusionco-extrusion ofof reinforcement.reinforcement.

Nevertheless,Finally, a future the challenge structural performanceis to investigate of the the systems process depends on a larger not onlyscale. on In mere this geometry respect, two but alsoscenarios on other are parameters,considered possible. like node Firstly, connectivity, to scale and up on the the entire integration system ofand reinforcement. to print in very With large the former,containers it must filled be with ensured suspension, in each node and that secondly, the materials to modularize bind monolithically the structures to each and other build and large that noconstructions separating layer from is smaller formed, components. as occasionally As observed the CiS in process the experiments. is believed Regarding to be most the efficientintegration in ofprefabrication, reinforcement, a modu one majorlarization advantage basedhere on the compared maximum to the transportation layered build-up size seems of conventionally to be the most 3D printedfeasible structures, fabrication is scenario. the possibility Nevertheless, to extrude particularly material in as line the with limestone the prevailing-sand suspension force trajectories. is very Thisrobust unique (compared feature to makes gel), this it possible process to would co-extrude be feasible reinforcement, for onsite such production as a continuous as well. fiber In that or steelcase cableslarger withinspaceframes the concrete would strand. be printed Initial in testsan onsite have factory, been carried and be out lifted and areinto currently place using being cranes. developed further (Figure 16b). 6.2. SiCFinally,, Potentials a future and challenge Challenges is to investigate the process on a larger scale. In this respect, two scenarios are considered possible. Firstly, to scale up the entire system and to print in very large containers filled The two main features which were explored in the SiC process were, firstly, the gradation of with suspension, and secondly, to modularize the structures and build large constructions from smaller concrete by locally injecting discrete displacement bodies, and secondly, the integration of components. As the CiS process is believed to be most efficient in prefabrication, a modularization based continuous channels within a solid building element. Both features offer a vast potential for practical on the maximum transportation size seems to be the most feasible fabrication scenario. Nevertheless, application in construction. In addition to the possibility of making buildings generally lighter, the particularly as the limestone-sand suspension is very robust (compared to gel), this process would be production of lighter prefabricated parts also offers considerable potential for saving energy during feasible for onsite production as well. In that case larger spaceframes would be printed in an onsite transportation to the building site. Besides saving weight, and hence resources, SiC structures also factory, and be lifted into place using cranes. facilitate the integration of additional component functionalities. For example, through the 6.2.integration SiC, Potentials of channels and Challenges and ducts, active cooling or heating of building elements can be facilitated. Moreover, integrated channels may be used for structural purposes, for example, for the integration of postThe-tensioning two main cables. features which were explored in the SiC process were, firstly, the gradation of concreteWith by regard locally to injecting the location discrete of displacement production, bodies, both prefabrication and secondly, and the integration in situ fabrication of continuous seem possible. In further development, however, research will concentrate on prefabrication. This is due to the fact that the release and reuse of the suspension from the hardened concrete can be better controlled in a factory setting.

Materials 2020, 13, 1093 15 of 17 channels within a solid building element. Both features oﬀer a vast potential for practical application in construction. In addition to the possibility of making buildings generally lighter, the production of lighter prefabricated parts also oﬀers considerable potential for saving energy during transportation to the building site. Besides saving weight, and hence resources, SiC structures also facilitate the integration of additional component functionalities. For example, through the integration of channels and ducts, active cooling or heating of building elements can be facilitated. Moreover, integrated channels may be used for structural purposes, for example, for the integration of post-tensioning cables. With regard to the location of production, both prefabrication and in situ fabrication seem possible. In further development, however, research will concentrate on prefabrication. This is due to the fact that the release and reuse of the suspension from the hardened concrete can be better controlled in a factory setting.

6.3. CiC, Potentials and Challenges The CiC principle fosters the fabrication of high-performance components using mostly lower grade materials. For example, a normal concrete can be locally reinforced by injecting high-performance concrete in exceedingly stressed areas. This shows potential for a variety of practical applications. For example, the load paths in a component could be traced using a higher-quality material, so that the load-bearing capacity of the component increases along these paths. In the case of walls, for example, these would be areas in the immediate vicinity of doors or window openings. In addition, this method could also be used to locally reinforce areas of direct load introduction, for example in the supports where beams rest on a wall. While the CiC experiments described in this paper essentially used the same concrete, merely with the addition of black pigments, the challenge in further investigations will be the use of concrete with vastly diﬀering mechanical properties. The challenge here will be maintaining their printing and bonding compatibility. In addition to using lightweight concrete as a support medium, the use of recycled concrete will also be subject to future investigation. In terms of structural design, currently predominantly continuous geometries have been printed into the formwork. This approach could be extended towards printing strategies in which the properties of a building element change gradually from property A to property B. For this, an injection strategy similar to the one described in the SiC approach will be investigated. Of all three process variations, the CiC process seems to be most likely to be implemented in situ, for example, by mounting a lightweight robot onto standard concrete formwork, in order to locally inject high performance mortar into cast-in place concrete constructions. However, here too there is great potential for prefabrication, especially in view of the fact that this process can be combined and hybridized with the SiC process.

7. Summary and Outlook In this paper, an overall concept for a novel, so-called Injection 3D Concrete Printing technology was presented, which is based on the injection of a fluid material into another fluid material with specific rheological properties. Three variations of this method were described and evaluated: A) Concrete in Suspension (CiS): injecting a fine grain concrete into a non-hardening suspension; B) Suspension in Concrete (SiC): injecting a non-hardening suspension into a fine grain concrete; and C) Concrete in Concrete (CiC): injecting a fine grain concrete with specific properties into a fine grain concrete with different properties. Each of these variations of the Injection 3D Concrete Printing concept was verified through numerous pre-studies as well as a final demonstration object for each approach. Within this framework a particular focus was directed to the scalability and economic feasibility of the 3D printing processes. For this, material combinations were developed which meet construction requirements in terms of economy and robustness. While each of the three variants is associated with specific potentials and challenges, there are characteristics in common between the different processes that need to be investigated in the further Materials 2020, 13, 1093 16 of 17

course of research. Most importantly, the fabrication process needs to be scaled from lab environment to real scale. In particular, the extruder technology needs to be upgraded to meet reliable and robust industrial standards, and to be able to extrude larger quantities of concrete without interruptions. Moreover, the digital geometry generation and path planning should be extended to include a structural analysis of the generated geometry, as well as an automated feasibility check of the robotic trajectory, in order to avoid intersections of the already printed geometries. The rheological properties (yield stress and viscosity) of the materials need to be specified to enable geometric precision of the printed objects. In addition, the amount of creep and shrinkage of the extruded material has to be investigated in future research. Another aspect for further investigation is precise control over the increasing volume during the printing progress: the more material is injected into the other, the more the volume increases. The mechanisms of the rising material level and the associated positional changes of the already injected material will have to be investigated more closely in future research. Following these initial proof of concept studies, in future research the structural properties will have to be quantified. For all versions of the I3DCP process, specific benchmarking, e.g., regarding the weight/performance ratio, must be developed and performed. The implementation of the process in the construction industry can only be realized in the medium term through quantified comparison. Finally, it should be stressed that the Injection 3D Concrete Printing method depends on a close interaction between the fabrication process and material parameters, and that these aspects cannot be considered separately. Hence, a close collaboration of material science, structural design, and process engineering is required to successfully advance the research.

8. Patents A patent is ﬁled under: N. Hack, D. Lowke, H. Kloft, Injection 3D Printing, DE 10 2019 105 596.2, 2019. https://www.ezn.de/ezn-patent/injection-3d-printing/.

Author Contributions: Conceptualization, N.H., H.K. and D.L.; Data curation, N.H. and I.D.; Formal analysis, I.D.; Investigation, N.H., L.B. and S.G.; Methodology, N.H., L.B., S.G., H.K. and D.L.; Project administration, N.H.; Resources, H.K., D.L.; Software, N.H., L.B. and S.G.; Supervision, N.H., L.B., S.G.; Validation, N.H. and I.D.; Visualization, N.H., L.B. and S.G.; Writing—original draft, N.H.; Writing—review & editing, N.H., I.D., L.B., S.G., H.K. and D.L. All authors have read and agreed to the published version of the manuscript. Funding: The Junior Professorship for Digital Building Fabrication is funded by the Gerhard and Karin Matthäi Stiftung. The fine grain concrete was sponsored by MC Bauchemie. The APC was funded by Publication fund of Technische Universität Braunschweig. This research received no additional external funding. Acknowledgments: The investigations of Section5 were conducted by the participants of the research seminar “Digital Building Fabrication Studio” at TU Braunschweig. The conceptual framework of Section2 as well as the experimental setup was provided by the authors. The students, who participated in the course and conducted the casstudies CiS and SiC were: Group 1, CiS: Annahita Meshkini, Jan Zöllner; Group 2, CiS: Da Xu, Sven Lammers, Simon Schröter; Group 3, SiC: Julia Böhnlein, Lamise Hamad, Thilo Schlinker; Group 4, hybrid of CiS andCiC: Mélusine Balke, Nina Trümper. The CiC experiments were conducted by the authors: Norman Hack, Leon Brohmann, Stefan Gantner and were supported by Phillip Rennen. Phillip Rennen also co-supervised the course as a tutor. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References

1. Bos, F.; Wolfs, R.; Ahmed, Z.; Salet, T. Additive Manufacturing Of Concrete In Construction: Potentials And Challenges Of 3D Concrete Printing. Virtual Phys. Prototyp. 2016, 11, 209–225. [CrossRef] 2. Lowke, D.; Enrico, D.; Perrot, A.; Weger, D.; Gehlen, C.; Dillenburger, B. Particle-Bed 3D Printing In Concrete Construction—Possibilities And Challenges. Cem. Concr. Res. 2018, 112, 60–65. [CrossRef] 3. Buswell, R.A.; de Silva, W.R.L.; Jones, S.Z.; Dirrenberger, J. 3D Printing Using Concrete Extrusion: A Roadmap For Research. Cem. Concr. Res. 2018, 112, 37–49. [CrossRef] Materials 2020, 13, 1093 17 of 17

4. Neudecker, S.; Bruns, C.; Gerbers, R.; Heyn, J.; Dietrich, F.; Dröder, K.; Raatz, A.; Kloft, H. A New Robotic Spray Technology For Generative Manufacturing Of Complex Concrete Structures Without Formwork. Procedia CIRP 2016, 43, 333–338. [CrossRef] 5. Sama, S.R.; Badamo, T.; Manogharan, G. Case Studies On Integrating 3D Sand-Printing Technology Into The Production Portfolio Of A Sand-Casting Foundry. Int. J. Met. 2020, 14, 12–24. [CrossRef] 6. Meibodi, M.A.; Jipa, A.; Giesecke, R.; Shammas, D.; Bernhard, M.; Leschok, M.; Graser, K.; Dillenburger, B. Smart Slab: Computational Design And Digital Fabrication Of A Lightweight Concrete Slab. In Proceedings of the 38th Annual Conference of the Association for Computer Aided Design in Architecture (ACADIA 2018), Mexico City, Mexico, 18–20 October 2018. 7. Dini, E. D-Shape. Available online: http://d-shape.com/ (accessed on 20 February 2020). 8. Khoshnevis, B.; Hwang, D.; Yao, K.-T.; Yah, Z. Mega-Scale Fabrication By Contour Crafting. Int. J. Ind. Syst. Eng. 2006, 1, 301–320. [CrossRef] 9. Lim, S.; Buswell, R.A.; Le, T.T.; Austin, S.A.; Gibb, A.G.F.; Thorpe, T. Developments In Construction-Scale Additive Manufacturing Processes. Autom. Constr. 2012, 21, 262–268. [CrossRef] 10. Cobod. Available online: https://cobod.com/ (accessed on 23 February 2020). 11. Apis Cor. Available online: https://www.apis-cor.com/ (accessed on 23 February 2020). 12. Winsun. Available online: http://www.winsun3d.com/En/About/ (accessed on 23 February 2020). 13. Xtree. Available online: https://www.xtreee.eu/ (accessed on 23 February 2020). 14. Mechtcherine, V.; Nerella, V.N. Incorporating Reinforcement in 3D-Printing with Concrete (In German). Beton Und Stahlbetonbau 2018, 113, 496–504. [CrossRef] 15. Asprone, D.; Menna, C.; Bos, F.P.; Salet, T.A.M.; Mata-Falcón, J.; Kaufmann, W. Rethinking Reinforcement For Digital Fabrication With Concrete. Cem. Concr. Res. 2018, 1.[CrossRef] 16. Dressler, I.; Freund, N.; Lowke, D. The Eﬀect Of Accelerator Dosage On Fresh Concrete Properties And On Interlayer Strength In Shotcrete 3D Printing. Materials 2020, 13, 374. [CrossRef][PubMed] 17. Kloft, H.; Hack, N.; Mainka, J.; Lowke, D. Large Scale 3D Concrete Printing: Basic Priciples Of 3D Concrete Printing. CPT Worldw. Constr. Print. Technol. 2019, 1, 28–35. 18. Valente, M.; Sibai, A.; Sambucci, M. Extrusion-Based Additive Manufacturing Of Concrete Products: Revolutionizing And Remodeling The Construction Industry. J. Compos. Sci. Compos. Sci. 2019, 3, 88. [CrossRef] 19. Laarman, J.; Novikov, P.; Jorkic, S. Anti-Gravity Additive Manufacturing. In Fabricate: Negotiating Design & Making; Gramazio, F., Kohler, M., Langenberg, S., Eds.; Gta-Verlag: Zürich, Switzerland, 2014. 20. Hack, N.; Lauer, W.V.; Gramazio, F.; Kohler, M. Mesh-Mould: Robotically Fabricated Spatial Meshes As Reinforced Concrete Formwok. In Fabricate: Negotiating Design & Making; Gramazio, F., Kohler, M., Langenberg, S., Eds.; Gta-Verlag: Zürich, Switzerland, 2014. 21. Johns, R.L.; Kilian, A.; Foley, N. Design Approaches Through Augmented Materiality And Embodied Computation BT—Robotic Fabrication. In Architecture, Art And Design 2014; McGee, W., de Leon, M.P., Eds.; Springer: Berlin/Heidelberg, Germany, 2014. 22. Hajash, K.; Sparrman, B.; Guberan, C.; Laucks, J.; Tibbits, S. Large-Scale Rapid Liquid Printing. 3D Print. Addit. Manuf. 2017, 4, 123–132. [CrossRef] 23. Rhoné, J.; Thomas, A. Soliquid. Available online: https://soliquid.io/ (accessed on 15 December 2019). 24. Wei, R.; Chee, S.; Tan, W.L.; Goh, W.H.; Amtsberg, F.; Dritsas, S. Robotic Fabrication. In Architecture, Art And Design 2018; Springer: Berlin/Heidelberg, Germany, 2019. 25. Hack, N.; Lowke, D.; Kloft, H. Injection 3D Printing. Available online: https://www.ezn.de/ezn-patent/ injection-3d-printing/ (accessed on 28 February 2020). 26. Universal Robots. UR 10. Available online: https://www.universal-robots.com/products/ (accessed on 23 February 2020). 27. Rhino 3D. Available online: https://www.rhino3d.com/en (accessed on 23 February 2020). 28. Soler, V. Robots. Available online: https://github.com/visose/Robots (accessed on 24 January 2020).

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).