Nonlinear Model Predictive Control PhD | Making Industrial Algorithms Easier to Maintain and Update

Nonlinear Model Predictive Control PhD | Making Industrial Algorithms Easier to Maintain and Update

Research cut-off: 6 November 2026

Quick Decision

Promising with material uncertainties. This is a 3.5-year EPSRC Doctoral Landscape Award (DLA) project at the University of Sheffield, supervised by Dr J Anthony Rossiter. It targets a meaningful control-engineering trade-off: keep nonlinear model predictive control (MPC) useful and robust while making the algorithm simpler to maintain and update. The stated funding package includes tuition, a UKRI-rate stipend and a research grant; applications close at 17:00 UK time on 13 January 2027. The Sheffield DLA process allows international applicants, but anticipates at most 10 international places across the competition and considers Home applicants first. Project entry · DLA application rules

The supervisor’s recent work gives this topic a credible predictive-control setting, and Sheffield has a broad control-and-systems research community. The main decision issue is that the project description does not identify a nonlinear plant, application domain, algorithm family, validation setup or success measures. Before investing in a proposal, contact the supervisor to establish what “simplifying” and “robust” will mean in this doctorate, and confirm the award terms for the applicant’s fee status.

Research Problem and Doctoral Work

What is unsolved and why it matters

MPC uses a model to predict how a system may respond to candidate control decisions, then chooses decisions over a moving horizon. Nonlinear MPC can represent richer dynamics and constraints, but the online optimisation may become computationally demanding. Reducing that burden can make a controller easier to run and maintain, yet simplification can also change prediction quality, constraint handling, closed-loop behaviour or robustness. A recent NMPC study illustrates one possible route—using data-derived bounds to restrict an optimisation search domain—but its evidence is a simulation in one autonomous-vehicle overtaking scenario, not a general result. Search-domain reduction study

Advertised work and plausible research choices

The project description says the work will propose approaches that preserve insightful and robust decision-making while keeping maintenance and updates cost-effective, with a specific focus on simplifying nonlinear MPC. It motivates the work with industrial systems whose operating scenarios change, where a complicated controller may be expensive to update and could be replaced by a lower-performing but easier-to-maintain alternative. These are the advertised aims; no particular system or method is named. Project entry

The likely doctoral contribution is an algorithm or design framework plus evidence about its trade-offs. Candidate choices to discuss—not confirmed project requirements—include which parts of the nonlinear model or online computation can be simplified, what information should remain explicit to support maintenance, and how to compare performance, constraint behaviour, robustness, computation and update effort. Sheffield’s wider control group works on MPC, dynamic optimisation, robust and adaptive control, learning-based control and real-time numerical optimisation, so the environment is thematically relevant. Control and Systems Engineering research theme

The available project text does not specify the target application, nonlinear model, baseline controller, software/hardware platform, data access, test rig, formal guarantees, industrial partner or measurable success criteria. Do not assume the project will adopt PFC, a specific solver or any application from the supervisor’s prior papers. Those points determine whether the 3.5-year scope is research-feasible and should be agreed early.

School, Programme and Field Context

The degree-awarding institution is the University of Sheffield; the project is listed in its School of Electrical and Electronic Engineering. Sheffield’s QS World University Rankings 2027 position is 82nd globally and 14th in the UK. Separately, its broad Faculty of Engineering submission ranked 8th for research quality in REF 2021, with 96% of activity in the top two quality categories and a top-national research-environment rating. The REF submission covered six engineering departments and the AMRC/Nuclear AMRC, so it is a faculty-level signal, not a ranking of this project or supervisor. Sheffield rankings · Engineering REF 2021 results

Sheffield describes full-time PhDs as normally lasting 3–4 years, depending on funding and scope; this project advertises 3.5 years of funding, which is within that range. The exact registration period and how thesis-writing time relates to the funded period should appear in the offer. The project is a Sheffield DLA studentship, not an advertised joint degree. The department-specific academic entry requirement is not stated in the project entry and should be confirmed with the school. The research-degree English table lists IELTS 6.5 overall with no component below 6.0, or equivalent, for Automatic Control and Systems Engineering and Electrical and Electronic Engineering. PhD application essentials · Research-degree English requirements · University location

Supervisor and Lab Assessment

Supervisor research directions

Dr J Anthony Rossiter’s profile identifies predictive control as a substantial research area and describes an interest in improving computational efficiency and simplicity while limiting performance loss. Sheffield’s control-and-systems theme includes MPC and dynamic optimisation, robustness, nonlinear systems and real-time numerical methods. The theme lists industrial and academic collaboration and a Rolls-Royce University Technology Centre in Control, Monitoring and Systems Engineering. The public information does not establish that this specific studentship is formally attached to the Centre, has a named industrial collaborator, or has access to a particular experimental plant. Rossiter profile · Control research theme

Recent important publications

• Rossiter et al. (2022), “A novel approach to PFC for nonlinear systems.” The paper proposes a computationally efficient predictive functional control (PFC) law for nonlinear systems with a simple tuning choice; selected simulations illustrate feedback and constraint handling. This is directly relevant to low-complexity predictive control, but PFC is not automatically the method for this PhD and the results are not a general robustness guarantee. Paper record
• Guerrero-Fernandez, González-Villarreal and Rossiter (2022), “Efficiency-aware nonlinear model-predictive control with real-time iteration scheme for wave energy converters.” This study applies real-time-iteration NMPC to a wave-energy control problem and derives a condensing method to reduce computation; it tests the approach on a scaled Wavestar design. It demonstrates relevant NMPC and computation experience, while remaining a specific application and scaled-model evaluation. Paper record
• Rossiter, Abdullah and Aftab (2024), “Improving the feedforward component for recent variants of Predictive Functional Control.” The paper combines feedforward information with newer PFC parameterisations to improve consistency and tuning. Evidence is principally simulation-based, and the authors identify sensitivity to uncertainty as an open question. This is a useful example of simplicity and tuning gains requiring careful robustness checks. Paper record
• Rossiter and Aftab (2025), “Efficient dual loop PFC for challenging dynamic processes.” The authors propose a dual-loop strategy for difficult, unstable or poorly damped process dynamics and test two industrial case studies in simulation. It indicates continuing work on simple predictive control, but is not evidence of industrial deployment or a guarantee for nonlinear MPC. Paper record
• Rossiter, Gonzalez-Villareal and Sarbini (2023), “A review of parameterised MPC algorithms.” This comparison proposes criteria for evaluating parameterised MPC and examines sampled low-order systems. The paper explicitly focuses on the linear case and points to nonlinear extensions as future work; it is a review/comparison rather than an implementation of this advertised project. Paper record

Taken together, these publications support a close thematic fit between the supervisor’s work and efficient predictive control. They also suggest that the PhD should define simplification alongside tests for stability, uncertainty sensitivity, constraint satisfaction and computation, rather than treating low computational cost alone as success.

Funding, Application, and Living Cost

The official project entry describes a fully funded 3.5-year award covering full tuition, a maintenance stipend at the UKRI rate and a research grant. It gives £21,805 per year as the current stipend. UKRI’s published 2026/27 minimum is also £21,805, with an indicative fee level of £5,238; the 2027/28 rates are not yet listed at this research cut-off. The published figure is therefore a current benchmark, not a guarantee of the rate after the October 2027 start. The project page does not state the research-grant amount or whether visa, travel, relocation or dependent costs are covered. Project entry · UKRI studentship rates

The DLA requires its own online form; an ordinary Sheffield PhD application cannot substitute for it. The form asks three questions, each with a 300-word limit. Applicants can select up to two relevant projects, and Sheffield strongly recommends using both preferences where appropriate. It recommends contacting the supervisor before applying. Select a 1 October 2027 start date and submit by 17:00 UK time on 13 January 2027. Shortlisted Home interviews are expected in mid-February 2027 and international interviews in late March 2027. The DLA recruitment statement also requires applicants to disclose and explain AI use in the application process; read the current rules before preparing or submitting materials. DLA application and AI-use information

International applicants may be considered, but EPSRC limits international recipients to 30% of places. Sheffield anticipates up to 10 international awards across the competition, assesses Home candidates first, and says it will fund the difference between Home and International tuition fees for successful international candidates. The project entry says most places go to Home-fee candidates. Fee-status classification and whether this particular project remains available after the Home phase should be verified; the “worldwide” label alone does not establish an international place. International selection and fee rules

For living costs, Sheffield’s current examples include monthly groceries at £155, transport at £73, going out at £60, takeaways/eating out at £45, clothes/shopping at £33, mobile service at £15 and subscriptions at £14. Its international-student guide gives average accommodation rent of £500–£800 per month and groceries at about £30–£45 a week. These are planning estimates rather than a complete budget, and 2027/28 prices may differ. As a simple scale comparison only, £21,805 divided by 12 is about £1,817 per month; after £500–£800 rent, roughly £1,017–£1,317 remains for food and other expenses. The actual award and costs should be checked in the offer year. Sheffield living-cost examples · International-student rent and grocery guidance

At the research cut-off GOV.UK lists a £1,171 monthly maintenance-funds requirement for Student visa applicants studying outside London, for up to nine months. Sheffield’s 2 October 2026 update says this will rise to £1,203 for applications submitted on or after 30 November 2026. This is a visa proof-of-funds threshold, not the amount the University estimates applicants will spend; it may change again before a 2027/28 visa application. GOV.UK visa funds · Sheffield visa-rule update

Applicant Fit and Preparation

The topic is best aligned with applicants who can show foundations in control engineering, nonlinear systems, dynamic modelling, optimisation or applied mathematics, and who are comfortable implementing and comparing algorithms. Programming and numerical-computation experience would be useful preparation, but the project entry does not list a specific language, degree title, grade threshold or programming package as a project-specific requirement. These are suggested preparation points, not stated admissions rules. Sheffield’s formal research-degree IELTS threshold is given above; the school should confirm the academic entry requirement for this project. Project entry · Research-degree English requirements

The DLA application asks candidates to explain their interest, how their skills support doctoral research, and an example of using skills to address a research question. Useful evidence could include a control or modelling project, a clear description of the question and evaluation, and a concise account of how computation or system changes affected the design. Avoid presenting a particular PFC or solver as required unless the supervisor confirms it. DLA form questions

Relevant career directions include control-systems engineering, process/automation engineering, systems integration and research or development roles in sectors such as aerospace, automotive, maritime, logistics, energy and manufacturing. Skills England’s occupational map describes work spanning system design, installation, testing, commissioning, maintenance and optimisation. EngineeringUK estimates 6.3 million UK engineering and technology roles, around 19% of jobs, but that broad figure is not specific to controls, PhD hiring or vacancy growth. Career prospects will depend on the applicant’s domain experience and ability to translate research methods into deployable systems. Control-systems engineer occupational map · Engineering and technology workforce briefing

Next Actions and Questions to Verify

Before applying

1. Ask Dr Rossiter what nonlinear system or application is currently envisaged, and whether a specific model, dataset, plant or collaborator is already available.
2. Ask what “simpler”, “robust” and “maintainable” mean in the selection brief, and which performance, constraint, stability, computation and update-cost measures will matter.
3. Confirm academic entry requirements, Home-fee classification and international eligibility for this particular project, including whether it is likely to remain open after the Home-candidate phase.
4. Request written confirmation of the 3.5-year award package: tuition coverage, 2027/28 stipend rate, research-grant amount, any international fee difference, and excluded costs.
5. Prepare the three DLA answers within their 300-word limits, select a second project only if it is genuinely relevant, and follow the stated AI-use disclosure rules.

At interview

1. Ask whether the project will develop new nonlinear MPC, simplify an existing controller, or compare several design families; the public description does not choose among these.
2. Ask how the work will test robustness, feasibility, constraints, model error, runtime and maintenance/update effort, and whether results are expected only in simulation or also on hardware/industrial data.
3. Confirm the supervisory team, meeting cadence, training, software/compute resources and access to any experimental or industrial environment.

Before accepting

1. Check that the offer letter confirms the degree-awarding institution, registration length, funding period, fee status, stipend, research support and the handling of thesis-writing time. Sheffield normally registers full-time PhDs for 3–4 years, while this advertises 3.5 years of funding.

Evidence and Sources

Project, DLA and university

• Sheffield PhD project directory
• EPSRC DLA application, selection and AI-use rules
• PhD application essentials
• Research-degree English requirements
• University rankings
• Faculty of Engineering REF 2021 results
• Sheffield location in the United Kingdom

Supervisor, research environment and papers

• Dr J Anthony Rossiter profile
• Control and Systems Engineering research theme
• A novel approach to PFC for nonlinear systems (2022)
• Efficiency-aware nonlinear MPC for wave energy converters (2022)
• Improving feedforward in recent PFC variants (2024)
• Efficient dual-loop PFC for challenging dynamic processes (2025)
• A review of parameterised MPC algorithms (2023)
• Nonlinear MPC search-domain reduction (2023)

Funding, living costs and careers

• UKRI studentship stipend and fee guidance
• Sheffield living-cost guidance
• International-student rent and grocery guidance
• Student visa financial requirement
• Sheffield update to student visa financial requirements
• Skills England control-systems engineer map
• EngineeringUK workforce briefing