For a student with a draft literature review or state-of-the-art chapter in an engineering thesis, in mechanical, electrical, civil, chemical or computer engineering. This page shows what that chapter has to do in engineering, which sources besides journal articles a reader expects, how a comparison table of existing methods works, and how the gap turns into a requirement for your own design or method. It quotes department and library guides from Germany, Sweden, Finland, Canada and the US. For other chapters, see thesis chapter feedback.
Esmond doesn't open your sources. From within your field, it points out where your use of them leaves a claim unsupported, where you cite a source in two conflicting ways, and other blind spots.
The chapter goes by several names: literature review, state of the art, background, related work. Whatever the name, it answers a practical question. What already exists that solves your problem, how well does it work, and under which conditions was that shown? The Physics-Enhanced Machine Learning group in TUM's Department of Computer Science fixes the chapter in the outline of the theses it supervises, as section two: “State of the art (three to four subsections, each for one of the related topics you need in the main part, 20-30 percent of the text)”. The same group describes what a master's student has to reach: “Master students have to "reach the frontier", i.e. thoroughly read the literature on their topic and know what the current limitations are.”
At KTH's Department of Engineering Design, the background study is part of what earns the degree project its credits: “To obtain the 30 credits for your thesis, you must include an extensive reference section that reflects the solid background study you have done before the actual project work is performed.”
What sets the engineering version apart is the end point. A reader looks for performance data under stated conditions, and for a gap that your own design, model or method will address. The Chair of Urban Water Systems Engineering at TUM puts it in its guidelines for theses: “Equally important, you will need to identify a gap in the literature that you intend to fill. This is how you justify your thesis, and it helps the reader to assess the importance of your work”. For the gap in general terms, see the page on the research gap; for length, see how long a literature review should be.
A hypothetical example of a commentYour text concludes that the Kalman filter approach suits this pack at low temperature, but your comparison table lists five estimation methods, all tested at 25 °C, and your requirement asks for accuracy down to -10 °C. Which reviewed source supports that conclusion?
A first comment on one part of your own draft, right away and without an account.
Get a free previewJournal articles are one source type among several. Aalto University's School of Engineering lists where information for a master's thesis can be found, including “articles in scientific journals, reviews, congress papers, theses and international societies of the discipline”. Each type supports a different kind of claim, and a reader checks whether you use it for the claim it can carry.
The MEng project report guidelines of Western University's mechanical engineering department describe the literature review in one line: “It should be comparative in nature and critical where appropriate.” In engineering, the comparison often ends in a table. The rows are existing methods, designs or technologies; the columns are the criteria your problem sets. The criteria come from your requirements, not from whatever each paper happens to report.
A made-up example: a thesis on estimating the state of charge of a battery pack for a vehicle that runs in winter. The review compares four families of estimators.
| Approach | What it needs | Known weak point | What to record from each source |
|---|---|---|---|
| Coulomb counting | A current sensor and a known starting charge | Error builds up with sensor offset; drifts without a reset | Test duration, current profile, how the start value was set |
| Open-circuit voltage lookup | Rest periods and a voltage-to-charge curve for the cell chemistry | Needs the cell at rest; a flat voltage curve, as in LFP cells, gives poor resolution | Cell chemistry, rest time, temperature |
| Equivalent-circuit model with Kalman filter | Model parameters identified for the cell; voltage and current measurement | Accuracy depends on parameters that shift with temperature and aging | Temperature range, how parameters were identified, error metric |
| Data-driven estimator | Training data that covers the operating range | Behavior outside the training range is unknown | Training and test conditions; whether test cells were held out |
A careful reader checks the right-hand column against each figure. A reported error of 1% says little if the paper doesn't state the temperature, current profile and cell age behind it. Two other problems show up in comparison tables: errors measured with different metrics in one column (root-mean-square error next to maximum error), and a cell left empty or filled from a source that tested something else. Both make the comparison look tidier than the evidence is.
The UBC Engineering Physics Project Lab names the opposite problem among the common issues it lists for final reports: “Not enough discussion describing existing and state of the art techniques or solutions.” A table without discussion of why the approaches differ leaves the reader to do that work.
In an engineering thesis, a gap is useful when it can be stated as a requirement that your design or method addresses: a quantity, a range of conditions, a limit. "No study has looked at state-of-charge estimation for electric buses" is a statement about the literature, and a reader can often disprove it with one search. "The reviewed estimators report errors below 2% between 15 and 35 °C; none were tested below 0 °C, and this pack operates down to -10 °C" names what is missing, why it matters for this project, and what the later chapters will have to show.
Keep the criteria of the comparison table and the criteria of your own evaluation the same, so the reader can see whether your result closes the gap you named. How those criteria turn into a test plan is on the engineering thesis methodology page, and how the final chapter returns to them is on the engineering report conclusion page. The project report page covers the chapters around the review.
| What they check | How it shows in a draft |
|---|---|
| Coverage | Recent work and the established baseline methods are both present; a standard method the field uses isn't missing |
| Conditions behind each number | Test conditions sit next to every performance figure you cite |
| Source type fits the claim | Performance claims rest on tested results, not on datasheets or patent text |
| Criteria traceable to requirements | Each column in the comparison table appears in your requirements, with a value |
| Your position | The chapter says which approach you build on and why you set the others aside |
| Gap as a requirement | A quantity, a range or a limit that the method and results chapters return to |
These fragments are made up. None of them comes from a real thesis.
Esmond evaluates every assignment before you submit. It never writes it.
Esmond is an AI tool that doesn't write for you. If your course doesn't allow any AI tools, don't use Esmond for that work. If you're unsure, ask your instructor or advisor.
In a state-of-the-art chapter, Esmond can flag, in the text you upload, claims that one approach is better without the conditions behind them, comparison tables whose criteria don't match your requirements, datasheet or patent values used as test results, and a gap that the later chapters never return to, if the file holds them. See the sample report for the form of the comments and how Esmond works for the steps.
Esmond reads what you upload. If the file holds the whole draft, it can set this chapter against the others; if it holds one chapter, the review stays within that chapter and may raise a point that another chapter already answers.
Esmond doesn't run a literature search for you, check whether a standard edition is current, or add text to the chapter. You decide what to change.
In the TUM group quoted above, it is the second section of the thesis, with three to four subsections, one for each related topic the main part needs. Inside it, one workable order is: the problem and the criteria that matter, the existing approaches grouped by type, a comparison table, and the gap stated as a requirement. Your department's template decides the chapter name and its place.
Apart from the outline of the TUM group quoted above, none of the guides quoted here prescribes one for the content. The parts that recur are a comparison of existing approaches on criteria taken from your requirements, the test conditions behind each figure, and a gap your project addresses.
Yes. The library session for capstone students at Simon Fraser University treats both as sources for engineering reviews. Use each for what it can show: a standard sets a test method or a limit, a patent shows that an approach has been claimed, and neither replaces a published test of performance.
Not by default. A systematic review follows a protocol with a documented search, eligibility criteria and screening, and some engineering fields, software engineering among them, publish reviews of that kind. A state-of-the-art chapter in a design thesis selects and compares the approaches relevant to your problem. If your supervisor asks for a systematic review, the search needs its own method description.
The TUM group quoted above puts the state of the art at 20 to 30 percent of the text. Other departments give other figures or none. Ask your supervisor, and check what your department's guidelines say.
Read on November 8, 2026.
Get your paper evaluated before you submit