Aeroengine Diagnosis Using a New Robust Gradient-like Methodology
Round 1
Reviewer 1 Report
1. The recent studies are not well covered in the Introduction. Model-based, data-based and information fusion methods are widely used for EHM, and more literatures, especially in recent decade, should be added.
2. Is it enough sensor measurements to compute component degradation? Although, the number is ten to ten, most of them are installed before the combustor. The hot section should be paid more attentions.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
It is known that a publication can have scientific value in the following cases:
1) It considers a new problem or proposes a new way to solve a known problem.
2) A comparative study of several known methods was carried out, as a result of which the areas of their preferred application were determined.
3) A well-known approach is applied to a new object and the features of such an application are analyzed.
4) A qualitative review of the problem and known methods for solving it is performed.
Unfortunately, these features are not included in this publication. The authors stated that they proposed a new method, the novelty of which is determined by: 1) The iterative nature of the solution search algorithm. 2) The fact that, in addition to the parameters of the technical condition of the engine components, the value of the turbine inlet temperature is calculated. In fact, none of these features are new. The solution of this problem using a nonlinear mathematical model is given in many works. And the calculation of such important non-measurable parameters as turbine temperature, thrust and stability margins of compressor cascades have long been among the tasks of the Gas Path Analysis. The considered object (turbofan engine) is also not new. Moreover, the authors artificially chose such a number of measured parameters that are not typical for normal operating conditions, as a result of which the number of measured parameters became equal to the number of the desired parameters of the technical condition of the engine components. Such a problem has a trivial solution, but does not correspond to practice. The review of the literature by the authors confirms that they are not sufficiently familiar with publications in the field of GPA. For an initial introduction, the following reviews can be recommended: 1) Mohammadreza Tahan, Elias Tsoutsanis, Masdi Muhammad, Z.A. Abdul Karim. Performance-based health monitoring, diagnostics and prognostics for condition-based maintenance of gas turbines: A review. – Applied Energy, 198, (2017), 122-144.2) Luca Marinai, Douglas Probert, Riti Singh. Prospects for aero gas-turbine diagnostics: a review. – Applied Energy 79 (2004) 109–126.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
Dear Authors,
In the paper the employment of aeroengine diagnosis methodology, that also apply for other, more general systems is presented. The manuscript contents are well organized, details are explained enough, and the results are clearly presented.
The high temperature measurements used for engine control and health monitoring are typically made using sensors located downstream of the turbine. In paper, this sensor is labelled T5t. It would be valuable if the comparison between computed T4t and measured T5t was also discussed.
I was able to find only minor flaws, like”
- Line 289 – “each” is repeated twice
- End of last not numbered line after line 357 – should be degradations.
The data availability statement would be welcomed.
Kind regards,
Reviewer
Author Response
Please, see the attachment
Author Response File: Author Response.pdf
Reviewer 4 Report
The paper attempts to present improvement in optimization technique for both diagnostic inference in a degraded gas turbine during operation as well as estimating the turbine inlet temperature. Generally, there is a lot of vagueness in the paper which would require clarification. Take note of technical comments below.
Typographical Comments
=================
Author affiliation for Jose Rodrigo, missing a comma
Technical Comments
=============
lines 122-129: It is not clear what the authors mean by default sensors. Perhaps referring to Table 1 would be relevant.
lines 258-376: It is not clear how the degradation was implanted and the degree of such. Without this, presented results lack context. Also, it is not clear if this is experimental based accelerated degradation study or from real operation; or just using PROSIS software.
lines 378-386: Same lack of clarity on the degradation condition. i.e. line 383, ... the same Ndegrad. = 10 degradations ... "what do you mean by this? what represents a degradation scenario?". Also, no clarity on how the performance of two applied optimisation technique are benchmarked. Complexity of a technique alone is not a basis of such.
lines 396-400: Same lack of clarity on the degradation condition. i.e. line 398, ... with incremental degradations smaller than 2% ...
lines 401-402: there is a general lack of clarity on adopted sensor locations. Just stating the number of sensor data is not sufficient.
lines 415-458: it is a standard process of adding Guassian noise to test robustness of analytical models. When applied, stating the mean alone is not sufficient. You need to include the standard deviation of the noise as well as how much standard deviation has been adopted until the technique becomes in-accurate. Without determining this threshold, it appears to be a hand picking exercise. As ab added benefit, this threshold can also be a useful metric for comparing the robustness of both optimisation technique.
lines 459-493: conclusion is misleading. Note the following:
- It is not clear what degradation was implanted and what the approach detected. Error bars or values say nothing
- It is clear turbine inlet temperature was estimated. However, both techniques arrived at similar accuracy, what was the benchmarking metric. How does noise affect both? What is the threshold limit? this was not achieved
- It is not clear what the sensor sets are. Very vaguely used in the paper.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors do not change formulations of novelty and conclusions to the paper. The corrections that they introduced concern some additional literature citations and additional explanations why we need to know turbine inlet temperature (TIT).
Therefore, all my notations for the previous variant of the paper are actual.
Publication of the paper will be useful as one more additional work on implementing the known methods of the engine health parameters determining. Therefore, I don't mind publishing it.
But I object to the authors' assertion that they were the first to propose a method for determining the turbine inlet temperature i (including using a non-linear model and taking into account changes in the technical state of components). In confirmation of the fact that other authors have been doing this for a long time, and not only at a steady-state, but also at transient modes, I can cite the following literary sources:
1. Armstrong, J., Simon, D. Constructing an Efficient Self-Tuning Aircraft Engine Model for Control and Health Management Applications. — NASA/TM—2012-217806. https://ntrs.nasa.gov/search.jsp?R=20130001700 2018-03-27T05:40:02+00:00Z.
2. Csank, J. and Connolly, J. W., “Model-Based Engine Control Architecture with an Extended Kalman Filter," AIAA 2016-1623, 2016. https://ntrs.nasa.gov/api/citations/20170000940/downloads/20170000940.pdf.
3. Zhiyuan Wei, Shuguang Zhang, Soheil Jafari, Theoklis Nikolaidis. Self-enhancing model-based control for active transient protection and thrust response improvement of gas turbine aero-engines. – Energy, 242 (2022) 123030. https://doi.org/10.1016/j.energy.2021.123030.
4. Yong WANG, Qian’gang ZHENG, Ziyan DU, Haibo ZHANG. Research on nonlinear model predictive control for turboshaft engines based on double engines torques matching. – Chinese Journal of Aeronautics, (2020), 33(2): 561–571. https://doi.org/10.1016/j.cja.2019.10.008.
Author Response
See the uploaded file
Author Response File: Author Response.pdf
Reviewer 4 Report
Revised manuscript provides clarification regarding the limits of degradation, type of noise added for robustness check , default sensors and use of gas turbine performance software to provided simulated degradation study of adopted gas turbine configuration. Proposed optimisation technique offers performance benefit viz-a-viz status quo with the added benefit of turbine inlet temperature estimation. Results would be useful to researchers in the field alike.
Author Response
See the uploaded file
Author Response File: Author Response.pdf
