Introduction

Process systems engineering (PSE) is field of research that has seen significant growth since the 1950s (Stephanopoulos and Reklaitis 2011). PSE is a ___domain that focuses on the design and operation of complex process or energy systems. More specifically, it focuses on the development and application of modelling and computational tools to simulate, design, control and optimise processes (Dimian et al. 2014). These tools are often developed based on representations of the underlying physiochemical and biochemical phenomena (Stephanopoulos and Reklaitis 2011). In the past 70 years, the PSE community has produced a range of process synthesis and optimisation tools such as heuristics (Seider et al. 2004), mathematical programming (Grossmann 1990), graphical-based (Linnhoff et al. 1982) and process graph (Friedler et al. 1992). On top of this, artificial intelligence (Venkatasubramanian 2019) and machine learning (Lee et al. 2018; Ning and You 2019) have recently been receiving increasing attention in PSE as well.

Research publications within the PSE community have a distinctive difference as compared to other research domains. For instance, publications on experimental and analytical research tend to focus more on the results obtained. This in turn moves the focus away from the way the research was conducted. In recent times, however, journals have developed new platforms in which researchers can publish their ways of performing experimental and analytical research. These platforms include MethodsX journals, appendices, e-supplements, etc. The purpose of this is to essentially provide researchers a platform to showcase novel ways of performing experimental and analytical research. Such practice is not the norm for the PSE community. Instead, PSE researchers often (if not always) place the development of tools as the focal point of their publications. In other words, papers published in the PSE community give high importance to the novelty of the tools developed. The tools presented in these papers are commonly referred to via terms such as approach, methodology, method, procedure or technique (AMMPT) (Foo 2009; Foo and Tan 2016).

Table 1 presents the AMMPT terms used in some notable and highly cited papers in the PSE community. For instance, Duran and Grossmann (1986) presented a tool to simultaneously optimise and perform heat integration for chemical processes. They used “approach”, “method” and “procedure” to describe the tool developed. Linnhoff and Ahmad (1990) presented a “methodology” to design near-optimal heat exchanger networks based on energy and capital costs. In addition to “methodology”, Linnhoff and Ahmad (1990) used terms such as “method”, “procedure” and “technique”. Shah (1996) developed mathematical techniques to solve crude oil supply scheduling problems in a refinery. The terms which Shah (1996) used were “technique”, “procedure” and “framework”. On the other hand, El-Halwagi et al. (2003) used the terms such as “approach”, “method” and “technique” to describe a tool developed for targeting minimum fresh resources for resource conservation networks. From Table 1, it is evident that most AMMPT terms are used arbitrarily and interchangeably to essentially describe the tools developed. However, each of these terms possesses inherent distinctions and must be used according to their intended definition. This brings forth an important research question—how can each AMMPT term be used appropriately to describe the PSE tools developed? Unfortunately, there is no framework that clearly answers such question and defines AMMPT terms for the usage in the PSE community.

Table 1 AMMPT terms used in most highly cited papers in PSE community
Full size table

In the past, AMMPT terminologies have been researched in the field of pedagogy. According to Merriam-Webster Dictionary, pedagogy is defined as the strategy and practice of teaching. AMMPT definitions help form essential and basic concepts for educators to conduct teaching. From here, an equivalence can be drawn between teaching pedagogy and PSE research fields. Both research fields heavily rely on the idea of developing tools and presenting them in a systematic way for the audience to understand and adapt.

As such, this paper is aimed at examining the definitions of each AMMPT term and developing a framework for using them systematically in PSE research. The remaining sections of this paper are as follows: “Bibliometric analysis of process systems engineering literature” section presents a bibliometric analysis on the use of AMMPT terms by previously published PSE works based on citation number, geographic ___location and key journals. Next, “Definitions for approach, methodology, method, procedure and technique” section explores the definitions for AMMPT and investigates how these definitions can be applied to formulate a conceptual framework for these terminologies in PSE research. “Conceptual AMMPT framework” section then presents a conceptual framework based upon the distinctions discussed in “Definitions for approach, methodology, method, procedure and technique” section. Following this, “Pedagogical examples” section explores several pedagogical examples to see how the conceptual framework in “Conceptual AMMPT framework” section could explicitly map AMMPT definitions to the PSE work presented in the past.

Bibliometric analysis of process systems engineering literature

For this section, a detailed bibliometric analysis is presented on the use of AMMPT terms in PSE literature. The bibliometric analysis was performed on a list consisting of PSE literature from the past 15 years (i.e., 2005–2019). This list was generated via Scopus database. Based on the literature list generated, the following trends were analysed:

  • Number of papers for each AMMPT term,

  • Key journals for each AMMPT term,

  • Citation numbers for each AMMPT term,

  • Geographic distribution for each AMMPT term.

Table 2 shows the keywords defined within the Scopus database search engine for each AMMPT term. In addition to this, keywords such as “experiment” and “process control” were excluded from the search. This is because this perspective is limited to the classification of PSE tools used specifically for process synthesis and optimisation. In this respect, the ___domain of experimental research and process control is beyond the scope of this analysis.

Table 2 Keywords, constraints and limits in Scopus Search for each AMMPT term
Full size table

Based on the constraints shown in Table 2, several results were obtained. It is important to note that the search results were generated in January of 2020. The search results show that 767, 310, 840, 249 and 362 PSE research papers have used “approach”, “methodology”, “method”, “procedure” and “technique”, respectively. In fact, Fig. 1 shows the number of Scopus-indexed PSE papers that have used AMMPT terms in their work, based on year. As shown, the distribution of PSE work using both “approach” and “method” terms is higher as compared to “methodology”, “procedure” and “technique”. This observation suggests that a large fraction of PSE researchers are more inclined to use terms such as “approach” and “method”. This could be because of the misconception that “approach”, “methodology” and “method” terms have the same meaning, while procedure and technique do not. Meanwhile, Fig. 2 shows the distribution of AMMPT terms within key PSE journals. Among these key journals, Computers & Chemical Engineering journal contains the highest number of AMMPT terms used as compared to the others. This is consistent as Computers & Chemical Engineering is a journal dedicated to PSE research. Moreover, Fig. 2 confirms the trend observed in Fig. 1, whereby “approach” and “method” are the terms used more frequently as compared to the remaining terms.

Fig. 1
figure 1

Number of papers for AMMPT terms

Full size image
Fig. 2
figure 2

Key journals for AMMPT terms

Full size image

The results were then analysed further based on the citations (Fig. 3). From Fig. 3, it can be seen that PSE publications which use terms such as “approach” and “method” generally receive higher number of citations. In addition to this, the number of times AMMPT terms have been used based on geographic spread can be seen in Fig. 4. As shown in Fig. 4, the AMMPT terms are used more in regions such as Europe, North America and Asia. This could be due to the large size of the PSE community working in these mentioned regions. In fact, these regions are known for very popular PSE-focused conferences such as the PSE Conference, PSE Asia Conference, ESCAPE Conference, PRES Conference, FOCAPD and FOCAPO conferences.

Fig. 3
figure 3

Citation overview for AMMPT terms

Full size image
Fig. 4
figure 4

Geographic distribution of AMMPT terms

Full size image

In summary, general bibliometric trends indicate that PSE researchers tend to favour terms such as “approach” and “method” when describing the tools developed. However, the usage of AMMPT terms varies across the years analysed. This does not mean that one term is better than the other, but it may simply reveal a lack of emphasis and consensus on the appropriate usage of AMMPT terms. Hence, there is a need to first effectively distinguish AMMPT terms and subsequently develop a framework which PSE researchers may use when writing papers.

Definitions for approach, methodology, method, procedure and technique

In the previous section, a bibliometric analysis was presented to detail the distribution of AMMPT terms in the PSE literature. From the analysis, it was found that there is limited consensus on the use of these terms. Hence, this section explores the definitions and differences between AMMPT terms, and based on these definitions, a conceptual framework is proposed in the next section.

Approach

According to Hofler (1983), the term “approach” is defined as the basic philosophy or belief concerning a given subject matter. It is a way or direction used to address a problem based on a set of assumptions. These assumptions can often influence the way in which practitioners orient themselves towards all aspects of their work (Hofler 1983). In other words, the approach plays a big role in determining how a problem would be solved. These assumptions originate from a collection of theories, concepts and working ideas, and they serve as a practitioner’s outlook towards addressing their problem.

An example of something that qualifies as an approach is the approach towards process design. Process design can be viewed via two approaches, namely hierarchical design and concurrent design (Andiappan 2017). Essentially, these two approaches provide distinctive philosophies to design a chemical process. For instance, hierarchical approaches consider each design task (i.e., synthesis, design and operation optimisation) in a hierarchical manner, while concurrent approaches consider them simultaneously. Another example is related to deterministic and nondeterministic (uncertainty-based) approaches. Deterministic approaches in PSE assume a single operating scenario with average values taken for operating conditions. Meanwhile, uncertainty-based approaches show an appreciation of nonsteady-state scenarios and expect a certain level of ambiguity in some data. Two approaches clearly illustrate the difference in philosophy.

Methodology

A methodology describes the general strategy to solve a problem. Methodologies are required in order to realise the approach defined previously. It functions as a guideline, allowing the practitioner to make choices within a certain set of rules/boundaries (The Open University 2015). By this definition, “methodology” is evidently synonymous with the term “framework”. Thus, both terms can be used interchangeably. Complementary to this definition, a methodology can be understood as a system of methods used with set rules or criteria. With a methodology, a suitable method can be chosen. However, it is worth noting that there are cases where a method can be chosen without the help of a framework. In this sense, the method is an obvious choice and does not require the presence of methodology or framework. Frameworks can often be the outcome of postgraduate dissertations in PSE. These frameworks are a result of organising all the tools developed during the dissertation period. There are several thesis dissertations that have proposed frameworks of such nature. For instance, Chemmangattuvalappil (2010) presented a thesis containing a general framework for integrating process and product design tools. Ng (2015) later extended this work by presenting a thesis on the development of novel integrated process and product design tools for biorefineries. It is worth mentioning that although the term “techniques” was used in Ng (2015), the focus of the thesis was on the collection of product design techniques within a framework. Like Chemmangattuvalappil (2010), the framework in Ng (2015) guides decision-makers to integrate process and product design tools. More recently, Ten (2017) presented a thesis to incorporate safety and health aspects into chemical product design. In this thesis, a framework was proposed to essentially guide decision-makers to choose suitable methods to incorporate health and safety assessments into chemical product design.

Method

A “method” entails how an approach will be practically implemented (Hofler 1983). It encompasses the way in which a research problem would be solved (Burnham 1999). For instance, a designer may decide to take a nondeterministic approach to solve a design problem. To implement this approach, stochastic programming can be chosen as a method. If a methodology was present, the methodology would consist of a guideline of which method to use (e.g., stochastic programming, robust optimisation, etc.) to solve a design problem. But as mentioned previously, it could be possible that certain PSE tools may not have a methodology, as a specific method may have been clearly identified for executing the approach.

Often, methodology and method can be confused for one another. However, there is a key distinction between methodology and method. If a practitioner engages with a method and follows it like a recipe regardless of the situation, then it remains as a method. If the method is not regarded as a formula but as a guideline, then it would be clearly classified as a methodology (The Open University 2015; Brookshier 2018).

Procedure

Every method has a procedure. A procedure is a sequence of techniques (Educational Research Techniques 2019). In other words, it is a series of techniques or actions conducted in a certain order. Procedures can be defined by the practitioner or can be in the form of built-in algorithms. Examples of built-in procedures in PSE may include (but not limited to) simplex algorithm (Dantzig 1990), gradient-based (Tawarmalani and Sahinidis 2002), branch-and-bound (Land and Doig 1960) and global optimisation algorithm (Horst and Tuy 1996).

Technique

This level refers to those specific activities practised by users that can be observed and measured (Burnham 1999). Essentially, a technique is an immediate step taken, which yields an immediate result (Hofler 1983; The Open University 2015). Moreover, a technique is often part of a procedure. In PSE, some example of techniques would be activities including collecting data, defining equipment, setting operating conditions, inputting data, solving a model, etc.

Conceptual AMMPT framework

Based on the descriptions of each AMMPT term, a conceptual framework can be formulated. As shown in Fig. 5, the conceptual framework is layered, starting with the approach at the highest level. Following this is the methodology or guideline to achieve the approach. Using the methodology, one (or more) method(s) can be chosen for implementation. Each method chosen has a procedure. The procedure contains a list of steps (otherwise known as techniques) that must be done to obtain an immediate result.

Fig. 5
figure 5

Conceptual framework for AMMPT classifications in PSE

Full size image

The conceptual framework in Fig. 5 is then explored further using three pedagogical examples in “Pedagogical examples” section. The first example explores AMMPT classifications for mathematical programming tools. The second example analyses classifications for graphical tools. Finally, the third example examines the AMMPT classifications for a recent paper found in the PSE literature.

Pedagogical examples

In this paper, three pedagogical examples are presented to illustrate the conceptual framework shown in Fig. 5. These three examples focus on the suitable use of AMMPT terms to describe activities undertaken to design process or energy systems. Details of each example can be found in the following subsections.

Pedagogical example 1

The first example is related to programming tools in general. The following table summarises how activities involved in programming tools can be classified via the conceptual framework presented in “Bibliometric analysis of process systems engineering literature” section;

As shown in Table 3, PSE tools may be developed via programming tools. These programming tools are often based on several philosophies. For instance, in terms of the process design philosophy, programming tools in PSE can be built using hierarchical approaches or concurrent approaches. It is worth noting that “approach” emphasises the philosophy, assumptions and beliefs undertaken. In this sense, mathematical programming tools in PSE could be classified as having deterministic or nondeterministic approaches. The key assumption here is on whether a given process design problem will be solved but assuming steady-state (deterministic) or dynamic (nondeterministic) operations.

Table 3 AMMPT classifications for programming tools
Full size table

As for “methodology”, programming tools in PSE may not necessarily have a framework or guideline to follow. In fact, most programming tools in PSE are developed using a fixed chosen method: mathematical programming. Examples of mathematical programming methods may include (but not limited to) stochastic programming, multiperiod programming and robust optimisation. The chosen mathematical programming method consists of a “procedure” to produce a mathematical model. The procedure consists of a series of actions known as “techniques”. These techniques may include actions that produce immediate results. Some examples include defining equipment/technology, collecting data on technologies, formulating mathematical equations, solving the model, etc. The mathematical model is then solved using a set of traditionally established search algorithms such as simplex, branch-and-bound and gradient-based. These algorithms are traditionally established search procedures that are used by solvers to determine the optimal solution for a mathematical model.

An example of a mathematical programming tool can be found in Yee et al. (1990). The model developed by Yee et al. (1990) is a classical mathematical programming tool used to simultaneously target area and energy in a heat network for a given process. Yee et al. (1990) developed this model to address the inherent weakness in hierarchical approaches for energy targeting. Essentially, Yee et al. (1990) chose the concurrent approach to solving the energy targeting problem. However, this example did not contain a methodology or framework which guided the user based on a set of assumptions or rules. In fact, Yee et al. (1990) evidently stated that their model is based on superstructure representation modelling and it eliminates the reliance on heuristics. This feature essentially underlines their contribution as a method. The example then proceeds to explain the procedure for developing the superstructure. The procedure contains several techniques that produce immediate outputs including fixing the number of stages, formulating the mathematical equations, defining the outlet temperatures as variables, etc.

Pedagogical example 2

The second example is related to graphical tools in general. Table 4 summarises how activities involved in graphical tools can be classified;

Table 4 AMMPT classifications for graphical tools
Full size table

Just like in Table 3, Table 4 begins with defining the approach taken. As mentioned previously, the approach represents the general philosophy when solving a process design problem. Unlike Pedagogical example 1, pinch analysis tools possess characteristics that classify them as methodology. This is evident as the outcome of pinch analysis tools may vary based on either decisions made by the decision-maker or the nature of the pinch problem addressed. In fact, graphical tools often lead to multiple possible solutions instead of one solution. An example of a graphical tool would be the carbon emission pinch analysis (CEPA). CEPA is a methodology that can be executed via the energy pinch planning diagram (EPPD) (Tan and Foo 2007) or the automated targeting model (ATM) (Lee et al. 2009). Each method that has a procedure consists of steps. These steps, by definition, are considered as techniques. In the case of EPPD, these techniques include tabulating energy sources and demand data and defining the carbon emission factors for each type of energy source considered. Based on this, the supply composite curve is created by combining individual supply curves. This is also performed for individual demand curves. After this, the supply composite curve is shifted until the first pinch point is determined. From here, the minimum amount of alternative energy sources needed to reduce carbon emissions will be determined (Tan and Foo 2007).

Pedagogical example 3

Example 3 focuses on a recent contribution from Andiappan et al. (2017). This contribution presented a Design Operability and Retrofit Analysis (DORA) framework for energy systems. Table 5 summarises how activities involved in the DORA framework can be classified. Table 5 starts by defining the approach undertaken by Andiappan et al. (2017). As shown, the contribution considered a concurrent approach to design an energy system. This essentially refers to considering (instead of isolating) operational issues during the design phase. Following this, DORA provides an evident guideline for decision-makers to follow in order to achieve the concurrent approach. The guideline may yield several possible solutions depending on the decisions made by the decision-makers. Such property allows us to classify DORA as a framework or methodology. The DORA methodology provides a combined mathematical programming and graphical method. Input–output modelling (IOM) is a mathematical programming method used to conduct disruption scenario analysis (DSA). Meanwhile, the graphical method essentially visualises the feasible operating range of the designed energy system.

Table 5 AMMPT Classifications for DORA framework
Full size table

IOM has a procedure prior to developing the mathematical model. The procedure entails the preparation of an input–output table. The input–output table is prepared via several techniques that give immediate outcomes. These include defining equipment, collecting conversion data and mass separations and setting inoperability scenarios. After this, a retrofit option will be recommended from the DORA framework. This option will then be evaluated by another technique known as benefit-to-cost ratio analysis to determine whether the benefits outweigh the cost of implementing such option.

Conclusion

This paper presented a conceptual framework for the suitable use of terminologies such as approach, methodology, method, procedure and technique (AMMPT) in process systems engineering (PSE) research. In this paper, the definitions of each terminology were discussed. Following this, the presented framework was analysed using three pedagogical examples. The first example considered classifying the use of the AMMPT terminologies for general programming tools. The second example classified the AMMPT terminologies for graphical tools. The last example used the presented framework to classify tasks in a previously published PSE tool to design energy systems. Publications trends have shown that AMMPT terms were frequently interchangeable in PSE research. This is because terms such as “approach”, “methodology” and “method” were often confused for one another. Thus, the proposed framework will offer guidance on choosing suitable AMMPT terms in future PSE work to accurately describe tools developed for process synthesis and optimisation.