Thermoforming of wood-plastic composites: a compolytics-approach translating combined polymer and policy analyses into industrial design principles

Wood-plastic composite (WPC) saves plastics, but products are still limited to linear decking and cladding. For advanced productions of three-dimensional WPCs, design principles were derived from seven published pre-studies on thermoforming. For this, a combined method of polymer research and socio-technological investigations reported in WPC research as compolytics-approach, derived a total effect-model for thermoforming and developed a decision tree with target group-specific settings of production parameters. Fourteen application-relevant material properties were influenced (p = 0.001) by thermoforming, with the strongest effects on colour (max. R2 = 0.93), followed by strength criteria (max. R2 = 0.41). Satisfying private deciders’ preferences for optimal façade appearance, a highest possible temperature should be applied for narrow bending under high wood content. Professionals value maximal strength, which demands a compound-independent wider bending at lower heat. The applicability of the design principles was assessed by case studies serving further research on WPC product development.


Introduction
Worldwide, 400 million tonnes of plastics are produced from petroleum, whereas plant-based polymers still hardly play a role with only 2.11 million tonnes [1]. This is because the price of bioplastics is 4 to 10 times more expensive or confidence in them is still low [2,3]. Sustainable alternatives are therefore needed that reduce dependence on fossil raw materials without significantly restricting consumption. Instead of banning petrochemical plastics, at least parts of them can be replaced by biomass [4]. Such substitution technologies already exist, embedding plant fibres into existing plastics [5]. Under the same volume and sufficient properties, many products can be made more sustainable. Wood-plastic composite (WPC) represents one such technology (Fig. 1). It consists of thermoplastic polymers in which up to 80% wood fibres are embedded [6]. The main applications today are in the construction industry with linearly extruded decking and cladding profiles [7] (Fig. 1a-c). Injection moulding is also used occasionally, for example as flat WPC floor tiles [8] (Fig. 1d). Examples of three-dimensionally shaped façade elements already exist, but only for other materials like fibre cement (e) and metal (f).
The development and production of WPC products are challenging because hydrophobic plastic and hydrophilic wood fibres need to be combined by means of adhesion promoters and desired properties to be created by additives [11,12]. WPC is therefore a compound whose formulation has been optimised for specific applications. Which properties these must ultimately fulfil in order to provide maximum utility to the consumer, is a socio-technological question besides material criteria [13]. For this, further analytical tools must be applied, such as discrete choice experiments or interviews with the company and private deciders [14]. Finally, the substitution of petroplastics should also relief scarce resources and avoid environmental damage from consumption and disposal. To employ WPC's maximal utility here as well, regulatory policy instruments must help shaping the sustainability transition [15]. Model-theoretical analyses find out how WPC distributes maximum gains equally among producers, consumers and the environment. This principle is known as Elkington's Triple Bottom Line (TBL) approach [16]. Such theories often provide the basis for Pareto-optimal policy design [17].
It seems plausible that product development with WPC is a holistic challenge for successful plastic transition. To this end, an approach emerged in recent years which, in the research literature, is reported as composite polymer and policy analytics, short -compolytics [18]. A compolyticsoriented development makes WPC maximally biobased so that it is still workable and functional. Furthermore, target applications are not only sustainable but also practicable and offer incentives to maintain a long-lasting establishment in industry and markets. Finally, this approach also generates niche potential for biomaterials as alternative to expensive bioplastics.
Under this philosophy, research outputs on durable WPCs in building applications have already been generated [19], the first application of WPC in packaging has been investigated [20], the switch from synthetic clothing to bioplastics has been analysed [21], and welfare effects from the standardisation of WPC application in consumer goods were assessed [22]. Recent compolytics-research focused on how WPCs can be processed into three-dimensional elements for shaped façades. Table 1 summarises studies on WPC thermoforming published so far, comprising two review papers, five papers about experimental studies, and one paper about target group preferences.  [8], right. Example of one-axis formed façade elements made of fibre cement (e) [9] and bi-axial made of steel sheets (f) [10]  RP.1 (Table 1) derives from the most recent literature the state of knowledge about heat-related effects on physical and mechanical WPC properties and it recommends effective parameters for further testing. It is reported that the change in WPC due to heat is strongly dependent on fibre content, with significant variances expected above 60% [30]. Exceeding 135 °C ( H Tem), WPC behaves in a markedly ductile manner, which influences mechanical parameters in particular [31]. According to the literature, the best way to find out which WPC parameters are most meaningful here is by applying test standards, but also by interviewing experts and consumers in the field of application of corresponding WPC products in façades [32]. From these review findings, the research plan suggests investigating hot-pressing of flat WPC elements prior to thermoforming based on two different compound formulations (type A, B) and that the conditioning temperature should be below and above 135 °C to generate variances. Also, RP.1 planned to gather the preferences of professional and private deciders (study E) in order to establish design criteria for thermoplastic WPC shaping according to target group interests.
Finally, study A investigated physical change effects from hot-pressing as an offline process, i.e. using already extruded WPC samples that were subsequently conditioned. It was found that with increasing heat, the samples became darker, provoking significant changes in L* (brightness) and C* (colour intensity). Weight loss and change on density occurred only in the upper-temperature plateau H Tem (> 130 °C). With moderate heat, the WPC surface initially hydrophobises, but under H Tem, wettability increased significantly. Temperature and heating duration provoked higher water absorption compound independently as a universal effect. As a result of high R 2 , colour measurements, water absorption and contact angle react most sensitively to the forming process as production control variables. Hence, a darker and more colour-intensive WPC-façade surface can be expected after thermoplastic forming, but it might also absorb more water.
Concerning mechanical behaviour, study B (Table 1) found that heat-induced strength reduction is greater with higher fibre content. In RP.1, an embrittlement and an associated increase in fracture energy were observed [33]. This was confirmed in study B under growing fibre content, but the rise is slowed down by increasing hot-forming temperature. Since Brinell hardness (HBW) is more dependent on fibre quality [34], no heat-induced increase was actually derived here for the more fibre-containing type A. The pull-through strength of fasteners did not show any universal reactions at all, which even revealed contrary behaviour in both types. Modulus of rupture (MOR) decreased significantly under heat, but stiffness (MOE) improved only slightly, making the former parameter a reliable control variable. MOR is thus most important, and thermoformed WPC façades show lower wind load resistance. RP.2 (Table 1) derived thermophysical and geometric principles from the literature and defined the questions investigated by pre-studies C and D. In contrast to RP.1, properties concerning curved WPC elements were now of interest. Paper RP.2 reports that the cooling of previously heated WPCs can be assessed from the response of the thermal conductivity λ [35]. According to this, an increase in λ also leads to longer cooling times or a higher temperature level is achieved with the same heat exposure. Recent papers, therefore, see decreasing λ-values as a way to reduce climate loads in buildings [36,37]. As far as a thermally induced length change (LCTE) of WPC is concerned, some papers see its decline under higher fibre contents [38,39], and strong heat would split polymer chains, which also lowers the length change potential [40]. The literature also provides interesting aspects on possible restoring forces (RET) of deformed thermoplastics [41,42], which has hardly been researched before for WPC. Tight bending could thus provoke severe re-deformation in later applications. How this is related to the fibre content and the hot-forming temperature was still an open question. Accordingly, elements would have to be overstretched and fastening points in the curvature area withstands only lower wind loads.
Finally, pre-study C ( Table 1) found LCTE to decrease with fibre content and more so under high-temperature plateau. This would make thermoplastically formed WPC façades less susceptible to deformation. As for λ, it revealed a higher value for denser compounds, in line with the literature. The polymer chain scission feared in RP.2 must obviously have occurred under H Tem conditions as well, because λ actually decreased and a lower final temperature was measured under the same time period. This was not necessarily due to a decrease in density as a result of loss in molecular mass. Thus, the cooling time under highforming heat was also shorter, from which an increase in the specific heat storage capacity could be deduced. This was particularly evident for the lower-fibre compound B. Practically, the cooling effect for interiors of such formed WPC façades is greater for compounds shaped under large polymer content and high hot-forming temperature. For production control, LCTE should serve as variable, whereas R 2 of the λ-prediction model is too low.
The last experimental study of paper D (Table 1) revealed that increasing fibre contents become decisive for the design of bent WPC shapes because then MOR curved and MOE curved are reduced the most in the curvature segment, which is to the detriment of fastenings. A large material thickness has a similarly negative effect because here cracking occurs during tight bending, especially under low hot-pressing temperatures L Tem (< 130 °C), and the bonding with the fibre is weakened. High heat counteracts this because possible cracks are healed again by the viscosity of the polymer. Paper D also confirmed the literature that high temperatures lower the memory effect (RET), which is universal for both compounds. Also, parts are more likely to remain deformed if they are thinner. For corresponding dimensionally stable façades, wide bending under high temperature is therefore advisable. The restoring force RET is also the best indicator for a consistent production process because it reliably reflects possible deviations with a sufficiently high R 2 .
The papers on studies A to D revealed change effects on 16 decisive material properties for façade applications. Adjusting all process parameters in such a way that each response variable shows the least change compared to the zero-sample does not appear to be very effective due to the large number of parameters. For example, an improvement of the restoring force (RET) under H Tem leads to a significant deterioration of water absorption (WA). Involving fewer parameters seems more reasonable, but this means that the consideration must be focussed on preferences of the target groups "professional" and "private" deciders. To this end, the socio-technological study E (Table 1) found that manufacturers rather base their decisions on test reports or approval documents, which make reported characteristic values trustworthy. Private homeowners would focus on visual items instead. In concrete terms, this means that a decision path for the target group "experts" focuses more on mechanical characteristics. These should then show little change from hot-forming under a certain setting of the process parameters in order to become pareto-optimal. The decision path of the households rather includes physical characteristics. In addition to these two main criteria, the highest possible bio-content in WPC ranked next for both target groups. Each path should therefore include the process parameter "fibre content" in the optimality analysis. Finally, study E also showed that a positive contribution of a WPC façade to building climate regulation was of secondary importance, even slightly more so among privates. The regression coefficients β, identified in the logistic regression analysis of study E, finally allowed a preference-related allocation of all 16 WPC properties to the two target groups.
Studies A to D represent the results from regression analyses describing the material change behaviour from hotpressing and forming, and assessing the sensitivity of WPC to this processing technology. This paper now builds on the individual results of these seven pre-studies and consolidates the respective partial explanatory models into a total model, on the basis of which general design aspects for thermoplastic WPC forming for façades are derived according to target groups. Whether this holistic compolytics-approach also delivers practicable results is then verified by means of case studies. They demonstrate the design and production of formed WPC façades as prototypes. It will also be shown how the forming technology for WPC according to Paper RP.2 (Table 1) can be meaningfully combined with phase change material (PCM) to optimise constructions in terms of building physics.
For the state-of-the-art, the study can provide information on a production technique for three-dimensional WPC products that is still hardly applied and thus supports future feasibility assessments. For industrial application, it determines relevant indicators for production quality control that react particularly sensitively to the forming process. For the industrial product management, it can guide the selection of marketing tools in a target group specific way to maximise the benefits of WPC as a plastic avoidance technology in the market. For research, it determines which material properties should be tested for the analysis of compounds in thermal forming and whether an early combination of technological and social criteria is generally useful in research. The organisation of the paper is as follows: Sect. 2 explains the method for setting up the total model and deriving the design principles. Section 3 presents the analytical results and practised cases and the last section draws a conclusion and points to the need for further research.

Conceptual foundations
All experimental studies A to D use the same WPC test material as shown in Table 2 (above). These are extruded profiles from the manufacturers Konsta [43] (Type A) and Fiberon [44] (Type B), with the same high-density polyethylene (HDPE) matrix, only the fibre type and its proportion makes the difference as test variable (Fib). Further process parameters are the hot-pressing temperature (Tem), the heating time (Dur), the pressing force (Pres) during forming and cooling and tightness of bending radius (Rad) ( Table 2, middle). In most cases, they are dichotomous and were derived on a trial basis with the test material. For example, under the minimum temperature and duration, deformation should just be possible and the maximum temperature and duration should still prevent material decomposition. Since both compounds are unequally dense due to different fibre contents, the limit values for the denser type A are higher. Type B contains the minimum wood content of 50% specified by the German Quality Association for Wood-based Materials (Qualitätsgemeinschaft Holzwerkstoffe e.v.) to be considered WPC [45], and type A, with 70% wood fibres, represents a maximum bio-content typical for WPC in outdoor applications. Finally, the 10.6 MPa pressure represents a force of several 100 tonnes that is still industrially applicable to medium-size façade elements. Thus, five material-and process-related parameters are included in the study. From the pre-studies, 16 response variables are consolidated for  Environmentally friendliness (4) Design aspect (4) Regulation: internal (4) external (4) Service: general (4) customised (4) Response variables Attitude the total model. According to Table 2 (middle), these come from the application-relevant categories "physical material behaviour" (study A), "mechanical behaviour" (study B), "thermophysical behaviour" (study C) and "geometric material behaviour" (study D). The compolytics-approach, illustrated by Fig. 2, describes the new analyses that extend the previous studies. As can be seen, the variables are now assigned to socio-technological criteria, which come from a discrete choice analysis on the decision-making process of private and professional customers towards WPC façades ( Table 2, below). Study E (Table 1) uses a logistic regression analysis to determine whether consent is more dependent on environmental aspects. If so, fibre-rich WPCs of type A with the best thermal insulation behaviour as per category "thermophysical" ( Table 2, middle) would then find approval. Alternatively, design aspects may also play a role. Then the hot-forming parameters according to Table 2 (middle) should positively influence the colour change behaviour and the dimensional stability from the category "physical" and "thermophysical". And finally, the respondents may also be interested in loadbearing capacity, which then concerns mechanical resistance from hot-forming. Thus, the qualitative categories of prestudy E can be assigned to the technical criteria of study A to D, which is a prerequisite for deriving a decision tree with design principles for WPC forming (Fig. 2, bottom right).

Quantitative analysis
As shown in Fig. 2, studies A to D contain parts or all elements of the following regression model: For the total model, all test results were first normalised taking the mean value of their zero-samples as reference. Thus, the normalised value indicates how much % the conditioned sample deviates from the virgin one. For the decision tree, it was additionally determined that a deviation in the direction of improved properties is expressed as a positive value. For example, an increase in water absorption, as a result of hot-pressing and subsequent shaping under (1) Change effects by thermoforming = ± Tem * Tem ± Dur * Dur ± Pres * Pres ± Fib * Fib ± Rad * Rad Fig. 2 Study design under compolytics-approach, combining technical and socio-technical analyses into a set of rules (policy) for designing thermoformed WPCs H Tem, would be disadvantageous for the façade application, which is illustrated by a negative percentage. With the normalised figures, the partial model was then set up again through regression analysis and the total model was afterwards calculated with the 1400 individual values (Fig. 2, below). This two-stage procedure was necessary because the partial analysis still has comparatively high-fit indices that allow universal statements about thermal forming. In comparison, the total model is based on more contrasting data sets, which weaken the model fit. The partial models were finally illustrated graphically as bar charts and then discussed qualitatively.

Socio-technological combination
The introduction already deduced from study E that private homeowners consider design aspects as most important for WPC façades, and company experts are first oriented towards proven technical performance. In second place is sustainability, and building energy efficiency ranked third which was similarly relevant for both groups.
The question now arises as to whether hot-pressing for thermal WPC forming could be carried out under moderate conditions in order to manufacture more economically. The introduction already raised concerns that this might not lead to Pareto-optimal results, at least not for both groups equally. But perhaps for one target group such a differentiation could be economically justified. To figure this out, this study now extends previous research by analysing the derived partial models target group-specifically focussing on the regression coefficients β and the error term ε. As result from the normalised test values, ε stands for the average deviation from the zero-samples and therefore indicates how much a WPC property improves or deteriorates through hot-pressing under moderate process parameters (L-Level: {Tem; Dur; Pres; Fib; Rad} = 0). The β-factors additionally reveal how the value then develops further when changing to H-level (= 1). The main interest of private deciders is to achieve the highest possible final value for physical effects. If this value is far above 1.0, the result is even better than in the zero-sample, otherwise worse. The difference to 1.0 therefore expresses a partial utility for the decider group. The dichotomized L-or H-levels in the setting of the process parameters then serve as a design policy for WPC forming by means of hot-press technology. For example, forming under very tight bending ( H Rad) could be pareto-optimal for L Fib under the setting H Tem, H Dur and H Pres. In addition to the physical (design) and mechanical (strength) categories, other sub-preferences are also opinion-forming according to study E. Thus, λ for "building physics" and fibre share for "sustainability" generate further partial utilities, which are added to the previous ones and the result is then averaged. For them, the optimal setting, for which the average total utility is maximal, was also derived in the further course of the decision path. With the help of these assignments, an Excel algorithm was programmed, which selected the most utilitymaximising combinations of process variables and associated β-factors, and always added a further product property, and then optimised the variables again iteratively. The result is a decision tree with main paths for private and professional deciders. Along each branch, optimised settings for process parameters are determined, which may well change when additional criteria are added. By specifying the mean total utility in the respective node of the tree, the first-best design criteria for WPC forming can be recommended as policy.

Design examples for practice
For the practical case studies, 145 mm wide WPC slats of types A and B were processed. They were taken from already extruded and commercially available profiles from the manufacturers, as specified in Table 2 (above), whereby Type A was a hollow profile and Type B a solid one. For the production of the test specimens, the top and bottom of the profiles were first planned to neutralise the original surface finishing by the manufacturer. Afterwards, a longitudinal cut was made to saw off a top and bottom cover layer in 6 mm thickness and then the saw-cut surface was planed again. Thus, the surfaces of the wood-plastic slats were of the same quality as the inner section of the previous boards. The test specimens were finally cut to length from the abstracted 5-6 mm thick slats, just like the samples in studies A to D (Fig. 3a). Subsequently, press moulds were pre-modelled. For this purpose, plastic water pipes were cut into partial radii and assembled into a relief according to drafts and glued at the joints (b). They were then mounted in a wooden cast for the production of a concrete mould tool (c). In the last step, this male tool was again placed in a wooden cast and the female tool was concreted on top of it. The result was two concrete tools into which the heated WPC element could later be inserted (d). Since WPCs are diverse and will not respond in exactly the same way, the derived design principles are to be understood as policies, as described in the introduction. The study initially practised them on two shapes that demonstrate thermal forming initially in one axis. In an industrial additive process, this is easier to combine with one-dimensional extrusion technology. Nevertheless, with the third case study, two-axis forming is also presented in order to initiate further development research. Table 3 shows the results of the partial regression analyses from the individual studies and the total model is shown at the bottom. Included are only significant regression coefficients (p ≤ 0.05) with the error terms and regressors representing relative changes. In addition, Figs. 4, 5, 6, and 7 illustrate the effects graphically, with hatched pattern for high fibre content ( H Fib: type A) and dark colour denoting extreme heating ( H Tem).

Total model for WPC thermoforming
As can be seen from Table 3 (section "Physical"), the five response variables, relevant for privates, change significantly by the hot-pressing parameters at the L-level, where ε reveals C*, Φ and WA to improve initially. With increasing heat ( H Tem), Φ and WA worsen then because their β Temfactors show a change with their sign in opposite direction of the arrow in the first column. The colour, on the other hand, improves with L* and C*. As discussed, mechanical characteristics were more relevant for professional deciders. Of these, however, only three criteria were at all sufficiently significant for hot-pressing, namely MOR, MOE and E Imp . Improvement effects are initially not to be expected at all under moderate process parameters (see ε). Under extreme temperature, only MOE enhances since β Tem , as a positive value, correlates with MOE↑ + . Thermophysical criteria had a secondary importance for both target groups. Here, thermal length change (α therm ) first leads to a 14%-worsening for the application (ε = 1.14), but then improves greatly (β Tem = − 0.67) and points towards α therm ↓ + . Concerning model quality R 2 and by comparing the column lengths in Figs. 4, 5 and 6, it can be seen for the WPC material that physical characteristic values react most sensitively to the process (max. R 2 = 0.93).
So far, the observations have not yet concerned the actual shaping of conditioned specimens, which only becomes apparent in the partial model on geometric criteria. Here, one can see a strong change in the ε-factors from bending under hot-forming with L Tem (Table 3_ "Geometric": ε < < 1.0). The increase in heat brings improvements throughout, as revealed by the signs of the β Tem -values in relation to the first column. However, it is advisable to keep the fibre content low (note: β Fib = 0 under L Fib) and to bend widely (β Rad = 0 under L Rad). The comparison between hatched and not-hatched columns in Figs. 4, 5, 6, and 7 shows that a maximum bio-basedness leads to a significant degradation in only three out of 15 cases (C*, MOR curved , MOE curved ), but also to relevant improvements in just two cases (Φ, E Imp ). Overall, this result is beneficial for a façade that is as sustainable as possible under hot-pressing and thermoforming. As far as the H Tem-effect is concerned, Fig. 6 underlines an outstanding improvement of LCTE, and Fig. 7 for RET. In summary, it can be said that hot-pressing alone has a partially enhancing effect on WPC, but that significant worsening also occurs during forming at the latest. Using the fibre content as a strategic variable makes less sense, rather the shaping temperature should be chosen intendedly, i.e. under extreme heat, if the main preferences of the respective target group improve as a result. Among the process Fig. 3 Production of moulds for making prototypes in the case study. a WPC slats. b Relief design as blueprint. c Concreting the tool. d Press forms Table 3 Partial-and total-model with normalised input variables and significant regression coefficients Effects (↑↓ + direction for improvement)  Table 3 shows the total model including all 16 variables. It is easy to see that hot-forming, even under moderate conditions, changes all essential material parameters of the façade by an average of 8% compared to the zero-condition (ε = 0.92), and that the bending radius has the strongest effect (β Rad = − 0.14), followed by the hotpressing temperature (β Tem = − 0.08). Nevertheless, this total model naturally has the lowest quality (R 2 = 0.06) and is hardly applicable for forecasting purposes, which makes a target group-specific assessment, based on the results of study E, all the more necessary. Figure 8 shows the decision tree with the partial utilities as %-improvements (positive values) or impairments (negatives) as a result from utility-maximising settings. If private homeowners only consider design criteria, then thermally formed WPC façades offer an improvement of 10.3% compared to the non-formed original material, e.g. extruded façade panels (Fig. 8, below: 1st Level). However, shaping must take place under extreme conditions and with wide bending. Nevertheless, the additional effort seems justified compared to the added value and the guiding policy is:

Decision tree for optimal setting of process parameters
For privates, always shape WPC façades under high heat and widely (Policy 1) However, if tight radii are formed, there is an average of 12.6% less utility compared to the original material, despite very hot forming. Maintaining a wide bending radius pays off in the further course because shaping remains advantageous when load-bearing capacity aspects are added, and corresponding façades offer an even greater average utility by 14.7% compared to a purely design-oriented consideration (Fig. 8: 2nd Level). Based on Table 3, this is due to the fact that, according to the model "Geometric", wide bending under H Tem brings an additional (0.12 + 0.08)/2 = 0.10 = 10% partial improvement for MOR curve and MOE curve . In the further course of the preference order, average utility worsens as a result of predominantly negative β Fib -values according to Table 3 ("Geometric") or Fig. 7 (hatched columns), although a 2% improvement is added again under building physics criteria. For this, however, wide bending must be maintained, otherwise tight radii always lead to worse material properties compared to the zero-state. Thus, for the target group of private homeowners, extreme conditioning with wide bending remains the first-best option.
For professional deciders, according to Fig. 8 (above), narrow bending and extreme conditions (0.0%) and wide bending under low heat and duration (− 0.8%) hardly differ in utility from the zero-state, as long as only load-bearing aspects are in focus. Thus, strong curvatures in the façade are not necessarily disadvantageous and can be realised via hotter conditioning without any detriment. The design principle is, therefore: For professionals, increase the temperature only with tighter bending (Policy 2) As far as a higher bio-share under L Tem is concerned, the β Fib = + 0.16 for E Imp would add significant value, but this is already cancelled out even under wide bending with β Fib = (− 0.13 − 0.14)/2 = − 0.135 for MOR/MOE curve (Fig. 7, hatched columns). Thus, with moderate conditioning, the result drops to − 3.6% on average (2nd Level). A change to extreme conditioning and the associated rise to only 0.0%, as a consequence from β Tem = (+ 0.12 + 0.08)/2 = + 0.10, has a balancing effect here, but with more energy consumption. Therefore, the design principle could be: For professionals, increase the temperature with rising fibre content (Policy 3) For the further course of the analysis (3rd Level), the same findings as in the lower part of Fig. 8 apply, in particular, a 2%-gain from the decrease in λ after forming under H Tem conditions. All in all, it remains to be said that thermoplastic forming of WPC façades should take group-specific preferences into account, because then, for professional deciders, wide-curved elements could well be produced under moderate conditions, which brings economic advantages from lower heating energy and faster production times. For narrow geometries, manufacturing can be flexibly converted to high heat. If design aspects are in the foreground, especially for private deciders, then very hot forming should always be carried out, but tight radii should also be avoided. If such façades are mass-produced and offered, then very hot forming should always be used as much as possible and, especially with tight radii, conditioning should be longer due to β Dur = − 0.21, which minimises restoring forces.

Setting the utility-maximal policy according to decision tree
From the previous preferences and associated utilitymaximum settings, design criteria for thermally formed WPC façades can now be derived, which are summarised  Table 4. In order for the forming of WPC to be energy efficient and fast, L Tem becomes decisive, but under wide bending. With a rather low fibre content, this then results in maximum utility (Table 4_ "Experts": Δ utility = − 0.8%). This optimal setting is now defined for the practical study as Case 1,x using type-B material and should be interesting for the experts according to Fig. 8 ("Strength": 1-st level), provided they only favour maximum load bearing. In order to guarantee a consistent appearance of the WPC elements for private deciders, only wide shapes and strong heat should be applied, which then generates almost constantly high total utility nearly independent of fibre (Table 4_ "Privates": Δ utility = + 10.3% and + 8.5%). As can be seen in Fig. 8 (bottom: 2nd Level), they even get additional utility from improved load-bearing capacity, but then only under low bio-content (Δ utility = + 14.7%). This first-best option is now defined as Case 1,y and again uses type-B material. In order to nevertheless also demonstrate tight bending, Case 2 was defined as optimal forming for experts under H Tem for the high-fibre type A, which obviously results in 0.0% change in utility. For an industrial mass application (Case 3 ), forming should be as hot as possible, regardless of the bending radii, which then gives a maximum loss of load-bearing capacity (Table 4_ "Standard Product"). This was again realised with type A (Δ utility = − 9.7%). Table 5 now summarises the cases as a basis for the practical study. From these specifications, the designs according to Fig. 9 and the reliefs were created as a blueprint for the production of the concrete press moulds.
For hot-pressing, the temperatures used in the studies for type A (160 °C) and type B (105 °C and 140 °C) and corresponding heating times according to Table 4 were applied. After removal from the hot-press, the workpieces were placed in the concrete moulds, manually pressed into the relief and cooled between the male and female tools for 120 s under slight pressing force exerted by a screw clamp. Figure 10 shows the result of thermally formed WPC elements of type B in individual and group arrangement. Partially, the original grooved structure from the profile manufacturer was retained, which represents a) and b) alternating with smooth but brushed surface. In the pre-studies A to C  (Table 1), type A already demonstrated good compatibility with high heat, which was also explained by its greater density. In contrast, type B already reacts very sensitively at 140 °C, which Fig. 10c,) now illustrates. This makes it clear that WPC hot-forming must depend on the respective compound formulation and is hard to standardise. Therefore, the temperature was reduced to 130 °C for further Case 1,y -elements. The difference between high and moderate hot-press temperatures is illustrated in Fig. 10e, where the Case 1,y -elements on the right under H Tem show a darker and more contrasting surface compared to Case 1,x under L Tem, which study A already revealed. For realising Case 1,x while maintaining the specifications from Table 5, most samples were conditioned under 110 °C and 240 s. However, Table 3 shows an aggravation of the memory effect (RET) for moderate hot-press temperatures (ε = 0.86). This then proved true in forming practice, as illustrated by Fig. 10d. After removal from the concrete tool, the curvature was partially reversed, which became evident in comparison to Case 1,y -conditioned elements. Nevertheless, it was also possible to produce a corresponding structure for the façade mock-up (Fig. 10e).

Case 1
Finally, Fig. 10f shows both types and temperature-dependent results in a group comparison.
Overall, the upper temperature plateau makes thermoplastic forming under wide bending very effective for low-fibre WPCs which confirms Policy 1. Similar to the x-variant, Case 1,y -conditions must also have resulted in the greatest possible preservation of mechanical material strengths, as demanded by professional deciders. However, energy consumption is then far higher, which makes the x-variant the more efficient alternative. Figure 11 shows the results for the high-fibre type A with black pigmented matrix, likewise used in the pre-study C and D. The conditioning followed Table 4 at 160 °C for 300 s. The relief of Case 2 contains a mix of narrow and wide bending, which now increases the design possibilities compared to Case 1 .

Case 2
In contrast to the former case with the low-fibre type B, the forming of A under 20% higher share and intense heat proved to be much more practicable, which confirms Policy 3. The material surface did increase in roughness, which was due to the escape of volatile components of the matrix. This was also confirmed by pre-study B. The high hot-pressing temperature should be advantageous for the strength. At least Fig. 7 confirms that MOR curved loses less utility with increasing heat. With β Tem = + 0.12  (Table 3), this is a significant contribution to the fracture strength in the curved area. Also, with H Tem, the element stiffness increases as a result of β Tem = + 0.08 for MOE curved (Fig. 7). This is particularly advantageous for the double-curved relief of Case 2 , because in the area of the curvature with r = 35 mm, the profile cannot support itself on the substructure, which puts all the more stress on the two neighbouring curves. Any wind loads can thus still be effectively dissipated. Regarding the design, Case 2 permitted on the one hand a horizontal and vertical layout (Fig. 11a, e), but also an arrangement with shifted wave peaks. Pragmatically, this was possible by rotating the relief 180° in the direction of laying (Fig. 11b, f). The Case 2 -specifications additionally fulfil the demand of private deciders for high-contrast surface appearance. At the same time, the extreme temperature not only makes the forming of small radii in particular very practical, which now also confirms Policy 2, but it additionally increases the load-bearing capacity at the curve summit and reduces existing memory effects. This was also proven by the practical execution, because the cooled workpieces demonstrated very good dimensional stability after removal from the tools.

Case 3
Case 3 , in contrast to the two previous variants, represents a double forming, which involves partly very wide bending. This was done under H Tem conditions and, according to Table 3, would be associated with a decrease in MOR and E Imp in flat areas, i.e. not very much in the interest of professional deciders. Therefore, Case 3 could serve low-cost mass production of corresponding WPC façade elements, where cheaper prices result from reduced load capacity. For the applied A-type WPC, pre-study C could not derive a detrimental rise in λ. However, there was an increase in material temperature and cooling time. This thermophysical effect could be useful to the profile from Case 3 because the stored thermal energy can activate incorporated phase change materials (PCM). Following study C's theoretical discussion, Case 3 was filled with paraffin now. The result is shown in Fig. 12d) as paraffin granulate and e) after a phase change at approx. 57 °C, i.e. as molten and subsequently hardened. According to a), two shell elements could then form a PCM-capsule together. In terms of design, the two-axial relief suggested a shingle installation, as  Fig. 12f). Here, the joints can be arranged in a fishbone pattern, where the overlap changes from row to row.
In conclusion, hot-pressing and thermal WPC forming is a suitable technology for biaxial shaping as well, which expands the range of applications even more than Case 1 and 2 were already able to do. This opens up a wide field of research that provides room for creative ideas and gives rise to further application research in façades.

Conclusions and outlook
This study investigated the effects from hot-pressing of WPC with subsequent thermoplastic forming for multidimensional façade elements. In this study, partial models from five pre-studies of different kinds of properties were consolidated into a total model under a combined socio-technological analysis. This should make previous statements  9 Design process for drafts: Blueprint from different radii (left), production of a relief (middle) and the shuttering (right) Fig. 10 Thermally formed WPC elements according to Case 1 : a with original groove structure from the manufacturer, b group-assembling, c surface destruction with excessive H Tem/ H Dur. d difference between L-and H-conditioning, e higher colour intensity C*/L* from H-conditions. f vertical arrangement in the façade mock-up about material behaviour under thermal forming more target group-oriented and more general. Another novelty is that the guidelines obtained were directly verified in case studies. The knowledge gathered can serve as a starting point for further application-oriented research aimed at developing products using this new technology.
As a main finding, WPC forming can only be standardised to a limited extent, which requires a high degree of orientation of this technology to the respective compound. Nevertheless, this option is suitable for both individual and mass production, and effects can be calculated in terms of direction and intensity. The memory behaviour was found to be critical, which can be mitigated by high heat or wide bending. Strong heat was beneficial for visual appearance because this makes surfaces darker and more colour-intensive. This is countered by a decrease in strength and water resistance. Thermophysically, this technology opens up new aspects for WPC in connection with PCMs. Both technologies can be effectively combined and implemented in designs. WPC is considered a sustainable material because it is a plastic substitution technology that helps saving petroleum. Hot-pressing and thermoplastic forming are thus ecologically desirable, even if this process consumes energy. However, one of the results of this research was that this can  Nevertheless, the studies failed to prove the universality of this technology for WPC. The dependence of the hot-pressing and forming parameters on the compound formulation is too great. Thus, this research also adheres to the overall conclusion that WPC is a material that is difficult to control and that formulations must take into account not only the target application, but also the processing. However, it also emphasises that a compolytics-approach indeed identifies benefits that remain obscured in a one-sided, technological view. In the medium term, further WPC research should integrate forming technology and investigate other applications, such as in automotive or everyday products, because a lot of petroplastics are consumed there. It is also recommended to follow up on study C, which suggests a high combination potential of WPC with PCM. There are also applications of additive 3D printing of FRP [46,47]. Future research should investigate synergies with thermoforming technology.