2.1. Integration of Process and Product Design
Green manufacturing encompasses a variety of aspects to be considered, from the design of the product to the synthesis of the production process. It is thus important to integrate process and product design aspects as an overall design problem. A hybrid design method that integrates mathematical modeling with heuristic approaches was proposed by Hostrup et al. [
28] to simultaneously synthesise a separation process and select the suitable solvent. Eden et al. [
29] presented a systematic approach that formulates the process and product design problem as two reverse formulations. The approach first identifies the process design targets by solving the product design problem, then synthesises the production process by solving the identified design targets. Papadopoulus and Linke 2006 [
30] presented a systematic framework to solve an integrated solvent design and process synthesis problem. The molecular clustering approach was utilised in the framework to identify a Pareto optimal solvent that fulfils both the product and process performance targets. An approach that integrates computer-aided molecular design (CAMD) techniques and reaction network flux analysis was presented by Hechinger et al. [
31]. The presented approach was utilised in identifying biofuels and the production pathways for the design of biofuels. Ng et al. [
32] proposed a systematic methodology for the synthesis of biochemicals in integrated biorefineries. Biochemicals that satisfy customer requirements were designed using CAMD techniques, while biomass conversion pathways that produce the designed biochemicals were determined using a superstructural mathematical optimisation approach. This methodology was later extended by Ng et al. [
33] for the design of a mixture in integrated biorefineries to address the product design challenge where the desired product properties cannot be fulfilled by a single-component chemical product. Meramo-Hurtado and González-Delgado [
34] presented a hierarchical approach for the synthesis and design of a multi-product biorefinery. Through decision-making, multi-objective optimisation, and numerical methods, an optimal biorefinery was synthesised by considering different feedstock and final products. The concept of decision-making was also incorporated into the work presented by Lai et al. [
35] to consider the cooperation among the corporate unit, business unit, research and development unit, and production unit in a chemical design and production enterprise. Restrepo-Flórez and Maravelias [
36] developed a superstructural framework for the design of biorefineries that utilises ethanol to produce gasoline, jet fuel, and diesel. The framework considers aspects such as catalysis, process synthesis, and fuel property modeling in designing biorefineries that produce fuels with specified product properties. Recently, Tey et al. [
37] proposed a comprehensive framework for the design of value-added pharmaceutical products from biomass. A chemical reaction pathway map (CRPM) was utilised to connect raw material, potential conversion pathways, and final products, while a mathematical optimisation approach was used to identify the compromised solution that considers the gross product and sustainability index. While these works discussed in detail the integration of process and product design, it is realised that the consideration of green manufacturing into the overall process and product design is limited.
2.2. Methodology
In order to incorporate green manufacturing into the overall product and process design, a systematic approach is developed to integrate product and process design with the measurement of green manufacturing. Definitions of green manufacturing in product and process design are first defined and categorised. This is followed by the design of an optimal green chemical product that fulfils the defined green and customer requirements by using the group-contribution method. Based on the designed product, an environmentally friendly conversion pathway is then determined by employing a superstructural mathematical optimisation approach. The step-by-step procedure to design a green chemical product and conversion pathway is presented as shown in
Figure 1. The developed approach shown in
Figure 1 can be separated into two stages, stage one that focuses on product design and stage two to identify the conversion pathway.
Step 1: Gather information for consistent green manufacturing definition.
The available green manufacturing definitions are reviewed and generalised into different categories. The purpose of generalisation is to group similar definitions that serve the same purpose and require the same green outcome on mankind and the environment. The generalised green definitions are further categorised in terms of product and process as shown in
Table 1 and
Table 2. This is imperative to allow easier identification of distinct differences between measurement methods required for the product and process. Additionally, categorisation provides comprehensive and clearer judgements on approaches to achieve green manufacturing needs. For instance, one of the green manufacturing product definitions is generalised as avoidance of unnecessary impact or use of energy after the end of usable life. This definition is needed and essential to be defined because the definition indicates the design of products that are able to break down into innocuous degradation products. Added to that, this definition restricts the design of high-complexity products that persist in the environment. By incorporating this definition into the initial product design strategy, end-of-life burdens of the designed chemical product can be significantly reduced.
Table 1 below shows a summary of the generalised green definitions for products.
It should be noted that most of the green definitions of products are focusing on the chemical industry. For other manufacturing industries, the information in
Table 1 should be updated to consider the important green performance indicators of the industry.
Table 2 below shows the list of generalised green manufacturing definitions for processes. One of the green manufacturing definitions is minimisation of waste and material consumption. This definition emphasises the process pathway to restrict waste generation instead of treatment for created waste. Decisions about the process pathway should be able to reduce the need for raw materials and incorporation of all materials used into the final desired product. Hence, this definition is essential to employ in the process design as it is the key to initiating practice to minimise the impact of undesirable output from processes on the environment and human health.
Similar to green definitions of products, green definitions of process, as shown in
Table 2, are developed for the chemical industry. While it is possible to utilise the definitions for other manufacturing industries, it is suggested to update
Table 2 for a more correlated and straightforward assessment of green performance indicators.
Step 2: Define objective for manufacturing process.
With the generalised definitions, measurement methods are first identified according to the target properties of the product to solve the product design problem. This is performed by identifying the appropriate green definitions of green products. Green products cover the targeted chemical and physical properties to ensure the designed product fulfils green manufacturing needs with reference to
Table 1. Apart from that, the properties of product requirement are identified to comply with customer needs. With these, the process pathway to produce the green product is determined with the definitions in
Table 2 to ensure both product and process fulfil defined green manufacturing. Hence, the objective is to develop a measurement approach that can meet green manufacturing and customer needs.
Step 3: Gather information on product.
Once the objectives for green products and processes have been identified, the defined green manufacturing and customer needs are translated into measurable properties. Method of measurement is identified to quantify the green manufacturing definitions. For example, Oral Rat LD50 can be used to ensure the designed green product is safe to use and will not cause a negative impact on the environment, which fulfils the definition of being harmless to human health and the environment. On top of that, the method of measurement and targeted property constraints are determined to guarantee that the product design solution achieves green manufacturing and customer needs. The property constraints are written as a set of property ranges bounded by upper and lower limits. The limits can be extracted from current environmental regulations and industrial standard. For example, the designed fuel product viscosity should fall between 0.30 cp to 0.60 cp to allow consistent fuel flow and lower pumping power requirements that fulfil customer needs. The properties constraints for product design are generalised as shown in Equation (1).
In Equation (1), represents the index for the target property. is the target value, while the constraints are represented as , the lower limit, and , the upper limit for the desired product property.
Step 4: Design product using inverse molecular design technique.
With the product information, the mathematical model using the group-contribution method is formulated to design a green chemical product. Molecular building blocks that are suitable for the product design problem are first determined in this step. The designed product with the combination of selected molecular building blocks should be able to replace the currently available product with similar functionality. For instance, CH- can be set as one of the molecular building blocks to represent the alkane functional group when the product design problem is to design a fuel product. Other than setting properties constraints in the model, as shown in step 3, structural constraints are also employed to allow the generation of a feasible chemical structure without the formation of free bonds. Assuming that only a single bond is considered, the structural constraint is illustrated as shown in Equation (2), based on Chemmangattuvalappil et al. [
52].
n1,
n2,
n3,
n4 refers to the number of degrees one, two, three, and four of a molecular building block that are available to bond with other molecular blocks, while
N represents the total number of molecular blocks in a molecule.
Moreover, the objective function is applied in the model to obtain an optimal solution of the product for a targeted property. For example, maximisation of a higher heating value can be set as the objective function when designing biofuel in order to fulfil engine efficiency according to customer needs. Subsequently, the solution for product design can be obtained. However, targeting on single objective is insufficient to provide a green chemical product solution. Other properties of the product will also be the key to designing a green chemical product. In order to consider multiple objectives simultaneously on several targeted properties, the chemical product design problem is solved as a multi-objective optimisation problem. The traditional weighted-sum method in solving the multi-objective optimisation problem requires decision-maker(s) to assign a weighing factor to each objective using expert judgement. In addition, the weighted-sum method might be biased as the weighing factors assigned to each objective are heavily dependent on expert knowledge or the personal preferences of the decision-maker. To address this, the fuzzy optimisation approach is incorporated into this mathematical model. The degree of satisfaction
λp of targeted property
p is introduced as shown in Equations (3) and (4).
λp ranges between the values of 0 to 1, which implies the level of satisfaction on targeted property value
Vp within the property constraints. A higher value of
λp indicates higher satisfaction of the targeted property. When minimisation of the property is required, the value of
λp approaches 1 when the obtained property value approaches the lower limit. Equation (3) is used when property needs to be minimised.
Equation (4) is utilised when the property need to be maximised. The
λp value approaches 1 when the obtained property value approaches the upper limit of the target property range.
The degree of satisfaction is split into three regions of above satisfactory, satisfactory, and below satisfactory, as shown in
Figure 2.
Next, the max–min aggregation approach is used to maximise the least-satisfied degree of satisfaction
λ. This is to ensure every targeted property is optimised simultaneously without any bias. To achieve this, the objective of the proposed method is to optimise the weakest/worst property among all targeted properties to be optimised. Thus, the least-satisfied
λ is maximised where the overall objective function is formulated as shown in Equations (5) and (6).
It can be seen from Equations (5) and (6) that the fuzzy optimisation approach identifies the relative importance of each targeted property to be optimised without the presence of a decision-maker, hence minimising the influence of bias on the chemical product design problem. To generate additional alternative feasible solutions, integer cuts are utilised. Integer cuts are applied as an additional constraint on the proposed mathematical model to avoid the same solution being generated again. The application of integer cuts can be continued until no feasible solution is obtained. Once the designed product achieved green and customer needs, Step 5 can be proceeded to determine the process that produces the designed product. If not, Step 3 needs to be repeated for further research to amend or improve product information.
Step 5: Gather information on process.
With the identified green definitions of process, the measurement method is determined to quantify green manufacturing needs on the process pathway. Information on all possible conversion pathways that convert specific feedstock to intermediates, then from intermediates into the final desired product is identified. With all the identified conversion pathways, the superstructure can be constructed. Next, parameters to measure the process performance can be identified. For example, the yield of the desired product is aimed to maximise the achievement of the green definition on minimisation of waste generation. To calculate the yield of the process, the conversion rate on each conversion pathway need to be obtained, which can be found through the literature. Other than that, the measurement of specific energy consumption can be formulated to acquire a process pathway that fulfils the minimisation of energy requirements and the green manufacturing needs. Specific energy consumption can be calculated from the heat of reaction through changes in heat of formation of each substance where the heat of formation of substances can be easily found. On top of that, all parameters are determined to guarantee that the process design solution achieves green manufacturing without neglecting customer needs. Customer needs can refer to the economic performance of the process, which is calculated as gross profit. The total capital and operating costs required for gross profit calculation can be obtained through references.
Step 6: Define objective function and formulate and solve the mathematical model.
Once the information on the process is gathered, a mathematical model can be created by formulating the design problem using a superstructural mathematical optimisation approach. In order to obtain the optimal solution to the conversion pathway, the objective function is applied to targeted parameters, which measure the performance of the process. For example, minimising environmental burden is set as the objective function when selecting conversion pathways. This is to fulfil green manufacturing needs where the prevention of pollution is considered when designing the process. Subsequently, the solution to conversion pathway selection to produce the desired product as found in the product design problem can be obtained. As multiple objectives are needed to be optimised, a fuzzy optimisation approach is utilised. With the developed model, an optimal process pathway that is green and economically feasible can be obtained based on the desired green product design.
Step 7: Configure the overall product/process design.
When green manufacturing needs on product and process are fulfilled, the developed mathematical model can successfully propose an optimal green product design with a green process pathway. If the solution to the product design does not achieve green manufacturing and customer needs, the steps cannot proceed to determine the optimal process pathway. This is because the desired product design is not identified. Hence, Step 5 has to be repeated to refine process information until the desired green product is achieved. After that, the refinement of process information is needed when the green process did not improve the feasibility of determining the optimal process pathway. The developed approach solves the process and product design problem as two design problems. Although an iterative feedback loop might be needed, this developed approach lowers the computational complexity of the problem. This reduces the computational efforts required to solve the problem compared with solving the integrated process and product design problem simultaneously. If it is required to solve an integrated process and product design problem simultaneously to obtain one optimum overall result, an algebraic approach presented by Bommareddy et al. [
53] can be utilised. However, simultaneous solution is not considered in this approach.
This work employs CAMD techniques and the fuzzy optimisation approach. It is noted that suitable property prediction models for product property estimation are not always available. Though data-driven techniques can be used in these situations, the accuracy of the identified correlation might be a concern to the user. In addition, it can be seen from Equations (3) and (4) that upper and lower limits for an objective are required in utilising the fuzzy optimisation approach, which might be challenging to obtain. Hence, this presented approach needs to be utilised judiciously by replacing the property prediction models and solution strategies depending on the nature of the design problem.
A case study is presented to show the efficacy of the developed two-stage optimisation approach.