Process Systems Engineering

Process systems engineering (PSE) is the branch of chemical engineering that treats a whole plant—or a whole company's chain of plants—as one system to be modelled, simulated, optimized, and controlled, rather than as a collection of separate vessels.

10 frameworks, from 1920 to today · Chemical Engineering · Engineering

still a working view ended · each leads to its entry below

Overview

Process systems engineering (PSE) is the branch of chemical engineering that treats a whole plant—or a whole company's chain of plants—as one system to be modelled, simulated, optimized, and controlled, rather than as a collection of separate vessels12. Its practitioners build mathematical representations of processes, use them to predict behaviour, and then search for the design or operating policy that does best against some goal: least cost, least energy, most product, fewest emissions, safest operation13. What is at stake is scale and coupling. A refinery or a pharmaceutical plant has thousands of interacting variables, and a change in one place—a recycle stream, a reactor temperature, a utility contract—ripples everywhere. PSE exists because intuition and hand calculation cannot track those ripples, and because the difference between a merely workable plant and a well-designed one is often enormous in money and energy.

From unit operations to the system as a whole

Chemical engineering's founding abstraction was the unit operation: the idea, crystallized in the early twentieth century and taught in the United States from the 1920s onward, that all chemical plants are built from a small set of recurring steps—distillation, absorption, heat exchange, filtration, reaction—each governed by the same physical principles regardless of the product being made. This was a powerful simplification, and it made chemical engineering teachable as a discipline. But it also encouraged engineers to design each unit well in isolation. The connections between units—recycles, heat integration, the way an upstream change forces a downstream adjustment—were handled by experience and rules of thumb.

The shift toward systems thinking came with the Second World War and the operations research that grew out of it. Linear programming, developed for military logistics, offered a way to optimize allocations subject to constraints. Chemical engineers recognized that refinery planning—which crude to buy, which products to make, how to blend—was exactly such a problem. By the 1950s, linear programming was being applied to refinery scheduling and blending, and the idea that a plant could be optimized rather than merely designed took hold. The term "process systems engineering" itself gained currency in the 1960s, as a generation of researchers began treating flowsheets as mathematical objects.

Flowsheeting: simulating the plant before building it

The central practical problem was this: given a proposed flowsheet—a diagram of units and the streams connecting them—what will the plant actually do? Hand calculation of a flowsheet with recycles is laborious and error-prone, because a recycle stream's composition depends on downstream results that depend on the recycle. The first general answer was sequential-modular flowsheeting, developed from the late 1950s and through the 1960s. Each unit is represented by a module with its own model; the simulator calculates units one at a time in the order the streams flow, and handles recycles by iterating—guessing the recycle composition, computing around the loop, and repeating until the guess and the result agree. Early programs such as the University of Michigan's work in the late 1950s and the commercial flowsheet simulators that followed made it possible to evaluate a design without building it. The approach's appeal was modularity: a new unit model could be added without disturbing the rest, and each module could use whatever specialized method suited it.

Sequential-modular simulation had a weakness. When a flowsheet is tightly recycled or when many specifications must be met simultaneously, the outer iteration converges slowly or not at all, and the engineer must choose which variables to "tear"—which streams to guess—a decision that could make or break the calculation. The competing approach, equation-oriented simulation, assembled all the unit models and connections into one large system of simultaneous algebraic equations and solved it directly with Newton-type methods. This handled recycles and coupled specifications naturally and converged fast when it converged, but it demanded good initial guesses and robust solvers, and it was less modular. The two approaches coexisted for decades and still do: modern commercial simulators typically offer both, with sequential-modular as the default for its reliability and equation-oriented mode for difficult, highly integrated problems.

Optimization: from flowsheet to superstructure

Simulation answers "what does this design do?" Optimization asks "what is the best design?" The decisive move was to stop optimizing a fixed flowsheet and instead optimize over a superstructure—a generalized flowsheet containing many alternative units and connections, from which the optimizer selects the best combination1. This idea, developed in the 1970s and 1980s, turned process synthesis into a mathematical program: choose which units to include and how to connect them, subject to mass and energy balances and cost. It made possible the systematic comparison of process alternatives that had previously been judged by experience. The approach's limitation is combinatorial: the number of possible structures grows explosively, and the resulting mixed-integer programs can be very hard4. Much of the field's later work has been about making these problems tractable and about combining mathematical synthesis with engineering insight.

Pinch analysis: thermodynamics as a design tool

A parallel and complementary line of work attacked energy. In the 1970s, following the oil shocks, Bodo Linnhoff and colleagues at what became the University of Manchester Institute of Science and Technology developed pinch analysis, a graphical and algebraic method for designing heat-exchanger networks. The insight is thermodynamic: plot the heat that must be supplied and removed across all process streams against temperature, and the curves reveal a "pinch" point where the process's heating and cooling demands are most tightly constrained5. The pinch sets a minimum achievable energy consumption for the plant, independent of the particular network built5. Engineers could then design networks that approach that minimum, recovering heat between hot and cold streams instead of buying fuel and cooling water5. Pinch analysis gave designers a target before they designed, and it spread widely into industrial practice and teaching. It also demonstrated a broader PSE habit: find the fundamental limit first, then close the gap to it.

Control: keeping the plant where the design put it

Design is only half the problem; the plant must then be held at its design conditions despite disturbances. Classical feedback control—a controller measuring an error and adjusting an input—was well established by mid-century, and chemical engineers adapted it to process loops. But a plant-wide problem emerged: with hundreds of loops, which variables should be controlled, by which manipulated variables, and how do the loops interact? Plant-wide control addressed this by designing the control structure as a system, deciding the hierarchy of loops and the pairing of measurements to valves so that the plant remains stable and operable as a whole. The question of controllability—whether a design can be controlled at all—became a design criterion, not an afterthought.

The most influential control development from PSE was model predictive control (MPC). Instead of reacting only to past error, MPC uses a model of the process to predict future behaviour over a horizon, computes the input sequence that best meets the objectives, applies the first move, and repeats6. It handles constraints explicitly—valves that cannot open further, temperatures that must not be exceeded—which classical controllers could not6. Early industrial MPC systems appeared in the 1970s and 1980s, notably in refining, and the approach became standard for multivariable process control. Its dependence on a good model ties it directly back to simulation: the same models built for design are reused for control.

Data, uncertainty, and the widening scope

From the 1990s, two pressures reshaped PSE. First, models were never perfect, and the field turned to data-driven and hybrid modelling: combining mechanistic models, built from physics and chemistry, with empirical models fitted to plant data, and quantifying uncertainty so that designs and controllers could be made robust to what was not known. Second, the scope widened from a single plant to the enterprise: planning, scheduling, and supply-chain decisions across multiple sites, integrated with process design and control. Enterprise-wide optimization treats the whole chain—raw material purchase, production, distribution—as one optimization problem, connecting the plant-level work of the earlier decades to business decisions.

The most recent turn is sustainability-oriented process design. Environmental constraints—carbon, water, waste—are now design objectives alongside cost, and PSE methods are applied to life-cycle assessment, carbon capture, renewable feedstocks, and circular processes. The same tools of simulation, superstructure optimization, pinch targeting, and model-based control are turned to minimizing emissions and resource use, and to designing processes that remain viable under uncertain future prices and regulations.

Where the field stands

Today PSE is defined by the integration of these strands: equation-oriented and sequential-modular simulation in commercial platforms, superstructure and mixed-integer optimization for synthesis, pinch and thermodynamic targeting for energy, MPC and plant-wide control structures for operation, and data-driven and hybrid models to bridge the gap between model and reality. The open questions are the ones the history keeps returning to—how to make optimization tractable as problems grow, how to trust models when data are scarce, how to design and control systems under deep uncertainty, and how to embed sustainability without losing the economic rigor that made the field useful. The discipline's enduring conviction, inherited from its origins, is that a chemical plant is best understood not as a pile of equipment but as a system whose behaviour can be predicted, and whose design can be chosen rather than inherited.

References

  1. Process engineering (introduction, Principal areas of focus in process engineering), Wikipedia
  2. Chemical engineering, Wikipedia
  3. Process optimization, Wikipedia
  4. Linear programming (Integer unknowns), Wikipedia
  5. Pinch analysis (introduction, History), Wikipedia
  6. Model predictive control, Wikipedia

The frameworks of Process Systems Engineering

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1920-1960

    Unit operation

    What common steps make up every chemical plant?

    Fly to its tower in Nousfarer

  2. 1958-Present

    Sequential-Modular Flowsheeting

    What will this proposed flowsheet actually do?

    Fly to its tower in Nousfarer

  3. 1965-Present

    Equation-Oriented Simulation

    How can all plant equations be solved together?

    Fly to its tower in Nousfarer

  4. 1975-Present

    Model Predictive Control

    How should inputs be chosen to keep a plant on target?

    Fly to its tower in Nousfarer

  5. 1975-Present

    Plant-wide Process Control

    How should all control loops be coordinated plant-wide?

    Fly to its tower in Nousfarer

  6. 1975-Present

    Superstructure Optimization and Process Synthesis

    Which process structure is the best of the alternatives?

    Fly to its tower in Nousfarer

  7. 1978-Present

    Pinch analysis

    What is the least energy a plant can use?

    Fly to its tower in Nousfarer

  8. 1990-Present

    Data-Driven and Hybrid Modelling

    How can models be improved with plant data?

    Fly to its tower in Nousfarer

  9. 1995-Present

    Enterprise-wide Optimization

    How should the whole company's chain be optimized?

    Fly to its tower in Nousfarer

  10. 2000-Present

    Sustainability-Oriented Process Design

    How should a process be designed to be sustainable?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Unit operationSequential-Modular FlowsheetingEquation-Oriented SimulationModel Predictive ControlPlant-wide Process ControlSuperstructure Optimization and Proce…Pinch analysisData-Driven and Hybrid ModellingEnterprise-wide OptimizationSustainability-Oriented Process Design

Beyond the field

Where Process Systems Engineering's ideas came from in other fields, and what they led to there: the strongest 12 of 46 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Process Systems Engineering is a planet in Nousfarer

Its city has a tower for each of these 10 frameworks. Land, walk in, and climb one concept at a time.

Land on Process Systems Engineering

Related fields

Fields its ideas are linked to

More in Chemical Engineering

See where Process Systems Engineering sits on the Map of Human Knowledge