For a student with a draft conclusion, or a discussion and conclusion, in an engineering thesis or technical report. It covers what the conclusion has to answer in engineering, how a result is stated with its numbers and the conditions it was tested under, how the limits of the test setup bound the claim, and how recommendations and future work follow from the results. Department guides from Germany, Finland and Canada are quoted. For other chapters, see thesis chapter feedback.
From within your field, Esmond points out where your conclusion goes further than your data, contradicts your results, or leaves out findings that don't fit.
An engineering conclusion is read against what the report promised at the start: objectives, requirements, a specification. Western University's mechanical engineering MEng guidelines tie the two together: “Conclusions must be based on the evidence presented and must be related to the stated objective of the project.” Aalto University's School of Engineering has a similar line among its evaluation criteria: “The thesis presents well-founded conclusions drawn from the results”.
That link starts in the introduction, where you state the purpose. At the end, the thesis guide of the Institute of Computer Science at Warsaw University of Technology asks the summary to “synthetically and objectively confront the results with the purpose of the thesis and confirm its complete (or incomplete) implementation.” If the goals in your introduction were measurable, the conclusion can say for each one whether it was met, partly met, not met, or not tested. For the conclusion in any field, see what goes in a thesis conclusion; the project report page shows where it sits in a full report.
If your report is a senior design project and not a thesis, the difference between a capstone project and a senior thesis and what is a senior thesis show how programs tell the two apart.
A hypothetical example of a commentYour conclusion states that the heat recovery unit meets the efficiency requirement, but your results chapter reports 81% at design flow against a requirement of 85%. Which test supports "meets"?
A first comment on one part of your own draft, right away and without an account.
Get a free previewWhere the report has a separate discussion, that section does the work the conclusion relies on. It compares the results with the requirements, with the published results from your state-of-the-art chapter, and with what your model predicted. It names the sources of error and says how large they are. The University of Waterloo's Chemical Engineering work-term report guidelines put the order plainly: “Conclusions are established based directly on your results and discussion.” For the split between reporting and interpreting, see results vs. discussion.
The UBC Engineering Physics Project Lab asks that the conclusions keep the numbers: “The Conclusions should provide precise information such as important numerical values from tests including the accuracy and repeatability of the measurements.” This matters most near a limit. If a requirement lies inside the expanded uncertainty of the result, the test can't decide on its own whether the requirement was met. The conclusion then has to say so, or name the decision rule it used.
A made-up example: a student tests a prototype heat recovery unit on a lab rig. Results are given with expanded uncertainty at k = 2. A table like this, in the discussion or just before the conclusion, lets a reader see each claim and its basis.
| Requirement | Result | Tested conditions | Status |
|---|---|---|---|
| Thermal efficiency at least 85% at design flow | 81 ± 2% | Design flow, 20 °C inlet air, lab rig | Not met |
| Pressure drop at most 150 Pa | 132 ± 6 Pa | Design flow only | Met at design flow; other flows not tested |
| No condensate at the outlet down to -5 °C inlet air | Not measured | Lowest inlet temperature on the rig: 5 °C | Not tested |
| Noise at most 35 dB(A) at 1 m | 34 ± 2 dB(A) | One fan speed | Within the uncertainty of the limit; not decided |
The Waterloo Chemical Engineering guidelines warn against claiming too much from one study: “Be realistic about your work: take pride in it but do not overstate its significance.” The Waterloo Civil and Environmental Engineering guidelines set the boundary: “The Conclusions should contain only statements that follow logically from the content of the report.” TUM's Chair of Urban Water Systems Engineering adds: “The conclusions must not contain any new content that was not already mentioned.”
In practice, each claim in an engineering conclusion carries a number and a condition. A result measured between 5 and 25 °C supports a statement about that range; anything outside it is an expectation and needs to be labeled as one. One prototype is one sample, so it says nothing yet about variation in production. A result that comes from simulation alone is a model prediction until a test confirms it, and the conclusion should keep the two apart.
The limits of the test setup belong next to the claims they limit. Lab instead of field conditions, short test duration (no aging), a narrow load or temperature range, one specimen, an instrument close to its resolution: for each, say which conclusion it bounds and in which direction. The general approach is on the limitations section page.
The Waterloo Civil and Environmental Engineering guidelines define the section briefly: “The Recommendations contain statements about additional work that might be required.” The Chemical Engineering guidelines set a limit: “Do not make broad recommendations that are not justified by the analysis and conclusions.”
Each recommendation in an engineering report points back to a result. In the example above, the efficiency shortfall leads to a design change to test (more plate area, a different flow arrangement), the untested condensate requirement leads to a test at low inlet temperatures, and the undecided noise result leads to a measurement with lower uncertainty or at more fan speeds. A recommendation can also be negative: not to proceed with a concept in its current form. "Future work could explore machine learning" points back to nothing in the report.
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 an engineering conclusion, Esmond can flag, in the text you upload, claims that go past the tested conditions, a requirement called met while the results chapter shows a different value (where the results are in the same file), results that appear for the first time in the conclusion, and recommendations that no result leads to. 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 rerun your tests or decide whether a requirement was met; it points to where the text and the results don't match. You decide what to change.
The status of each objective or requirement, with the measured values and the conditions behind them, and the main limits of the tests. In the Waterloo and Western guidelines quoted above, conclusions rest on the results and discussion already in the report and add no new results. Recommendations, if the report has them, follow from those conclusions.
The discussion interprets the results: it compares them with requirements, published work and model predictions, and weighs the sources of error. The conclusion states what follows from that interpretation, objective by objective, in fewer words and without new material.
In some departments, yes. The Waterloo Civil and Environmental Engineering guidelines allow the two to be combined, and the Chemical Engineering guidelines leave the choice to the student depending on the work. Check what your department's template expects.
Yes. Western University's mechanical engineering MEng guidelines ask students to give the reasons when project objectives were not achieved. A requirement left out of the conclusion is something a reader notices when comparing the conclusion with the introduction.
Waterloo's Civil and Environmental Engineering guidelines for work-term reports limit the conclusions to one page and prefer a single one-page section for conclusions and recommendations together. A thesis conclusion can be longer; your department sets the frame.
Read on November 8, 2026.
Get your paper evaluated before you submit