I find more than 10 % difference between the calculated results and the experimental results. Is this okay?
Predictions in structural engineering are not very accurate. For example, 20 % deviation in predicting failure loads of reinforced concrete beams is normal. 50 % Deviation in predicting crack widths in reinforced concrete is normal. Ask an expert in your field what deviations can be expected. If you obtain a better accuracy than 10 %, be prepared for critical questions because one might think that you fine tuned the parameters of your model to obtain this nice result. In that case the calculation is not a prediction but an "after diction". Parameter calibration can be useful but it should be clearly mentioned.
The mathematical model seems to predict the experimental results well. However, how can I proof this?
I would use a Monte Carlo Analysis. It is not difficult but you need to do the reliability course CT4130 to understand what you are doing.
I want to copy part of a book into my report. Is that okay?
From a scientific point of view the answer is yes, provided that you make a reference to this book. It is not sufficient to include just a list of all sources used. It needs to be clear which part of the text and which figures come from which source. Do not copy a reference to a source that you did not read yourself. So, if you copied part of the book of Dr. Smith than refer to the book of Dr. Smith and do not copy the references of Dr. Smith to books that you did not read. The idea behind it is to give credit where credit is due. Without this nobody would want to be a scientist.
From a legal point of view the answer is no. The copyright laws are strict and the claims that the lawyers make are even stricter. You can only copy a line or a figure when you have written permission from the copyright owner. The copyright owner is the publisher of the book. For example, officially you need a permission letter from the CEN if you write a formula from the eurocode in your design calculation. That would be unpractical and on their website the CEN tells us that they forgive us if we sometimes copy something small. You need to decide yourself. Keep in mind that they want to make a lot of money. If you interfere with that the lawyers will be send. That can be very expensive for you. Some people also lost their job, for example prof. Diekstra at Leiden University. If you decide to violate copyright then you better not refer to the source because that will make it easy for the lawyers to find you. They use Google too.
However, the Dutch copyright law makes an important exception for scientific publications. You do not need permission for copying into a scientific publication. So, if you live in the Netherlands and your report is scientific you can copy anything into it and put it on the WWW. There are some conditions, for example you need to refer to the source and the authors (See Auteurswet, Art. 15a, in Dutch). Other countries might have similar exceptions.
How can we interpret a correlation coefficient?
For example, many experiments on wood specimens have been performed. The information published is the mean and standard deviation of the modulus of elasticity, the tensile strength and their correlation coefficient. The detailed test results are often lost. A small Maple program can be used to see how the test data must have looked like. Run the program and vary the correlation coefficient to see the consequences.
How can I apply the Stringer-Panel method?
How many digits after the comma do I need to use?
Use 4 significant digits in your calculations and round the results to 3 significant digits. For example, (0.2340+13.58)*13.01-99.91/(0.6580-0.04590)=16.5 . Then you can expect a calculation error of about 1%. For example, (0.23395+13.575)*13.005-99.914/(0.65795-0.045904)=16.33. (16.5-16.33)/16.33*100%= 1.04%.
What font shall I use?
Choose the font yourself. However, it is common practice to make variables italic and units not italic. For example pmax = 30 N/m2.
What to do with peak moments and peak stresses in finite element results?
Peak moments and peak stresses computed with plate or shell elements are not always physically realistic. Often the peaks can be redistributed over some length. Some rules for this can be found in this Dutch text.
Shall I develop my model to be always on the safe side?
Please don't. Safety is accounted for by the partial factors that are used in design. Instead, try to make your model as accurate as possible. If you will do a reliability analysis the deviations will be included in the standard deviation of a model factor.
My supervisor wants me to fiddle with the analysis results so that they support his conclusions. What shall I do?
You probably have no choice than to follow his instructions, however, do not think that this is normal. Normal is presenting all relevant measurements and not only a biased selection. Normal is computing numerical results and not making them up. Normal is performing many checks. Normal is reporting everything necessary for others to obtain the same results.
If you feel comfortable with fiddling results than you need to consider a different career than in science. In science you could make a living but you would never discover something that future generations want to know. Just like your supervisor.
I am programming a new material model in a finite element program. The program gives strains and wants stresses in the integration point directions. However, the material model is in the principal directions. How can compute the transformation?
Use this code. principal values
How do I calculate torsion stresses in a I-section?
It is not as difficult as often suggested. Here is an example.
I want to become a Ph.D. student. What can I do to improve my chances?
Get a mark 8 on average (on a scale from 0 to 10), choose a graduation project that has a fundamental nature and be flexible in the subject you want to do research on.
Where can I get the eurocode?
Some organisations try making money by selling copies of the eurocode. Somehow, they obtained the copyrights from the scientists that worked hard to develop the code. In the USA this would not have happened because there one cannot obtain copyright of the work of a government employed person. Clearly, many people in Europe do not like paying a lot of money for codes that have been developed with their tax euros. I cannot give a link to the codes that I have because the lawyers check this page too. Surely, it is possible to download it somewhere free of charge. Paying for it would only make the lawyers stronger. If you use a Delft University computer and go to http://connect.nen.nl/ website you can download many codes for free.
Is there a formula for the buckling load of a column with varying cross-section?
No, but you can calculate the buckling load using potential energy. Here is an example.
Safety indices b are specified in the eurocode. So, every structure needs to have a b larger than the specified value. However, when partial safety factors are calculated the specified b is used as a target value. So, structures designed with these safety factors have the specified b on average but many of these structures have a b less than specified. What is the reason of this conflict in the eurocode?
I think this can be explained by splitting the people that develop the eurocode in two groups. Group B are the people that calculate safety factors. Group A are the people that do not. Group A consists of thousands of researchers. They are aware of the technical consequences if their formulas and factors err on the wrong side. So, they make safe decisions. They do not know how these decisions increase the costs of structures. Group B consists of a few dozens of experts. They have well developed calculation methods but wish they had more data on hidden safety in formulas and factors. They know that the construction industry will not accept large safety factors due to the costs involved. So, they came up with this ugly compromise to interpret the specified minimal b as a target value.
When checking 3D stresses in concrete we require that 1) all principal stresses are smaller than the tensile strength and 2) the Mohr-Coulomb criterion is fulfilled. Why are different tensile strengths used in the first and the second criterion?
The first criterion checks tensile failure. In this criterion the tensile strength is zero because a shrinkage crack might occur at the spot that we are checking and than the concrete has no tensile strength. The second criterion checks compression + shear failure. In this criterion the tensile strength has its uncracked value because this strength does not depend on shrinkage cracks. A shrinkage crack would be closed by the compressive force.
I have come up with a complicated model and implemented it perfectly ... but it does not work. What can I do?
Doing everything perfectly is not a successful approach when trying something new. You should have started with a simple model and made this one work. Subsequently, you should have made the simple model more complicated in small steps, making sure that each of these models produces reasonable results. The end result would have been a complicated model that does work.
How can I obtain many references to my paper?
Obviously, by writing a good paper on a scientifically popular subject and publishing it in a quality journal. However, the very best paper will be the last word upon a subject. It will be used by engineers and not referred to by scientists. Perhaps you should try to obtain very few references.
What are the conclusions to my project?
For good engineering research you need to split yourself in two; a person who wants to know stuff and a person who wants to build stuff. The "know person" starts the project, formulates the objectives or opportunities, writes the introduction, does the work and understands. About half way the project the "build person" cannot be ignored anymore. He or she asks questions such as; "Can I conclude this because knowing this would really help to build stuff?". The answer is usually, "No." The "build person" is not satisfied with this answer. The discussions are intense. It leads to changes in the planning. Also the objectives are rewritten. Finally, the "build person" writes the conclusions to your report. The "know person" is only allowed to check.
How will the grade of my MSc project be determined?
At the evaluation meeting each supervisor puts forward his or her grade. Supervisors who are not experienced in grading Delft projects give an evaluation in words. There is discussion and usually they agree to give the average. If there is no agreement than the chairman decides. The table below shows the grade distribution according to my experience.
| 1 | 3 | 20 | 26 | 30 | 15 | 4 | 1 | 0 |
| 6 | 6.5 | 7 | 7.5 | 8 | 8.5 | 9 | 9.5 | 10 |
My supervisors say that my communication needs improving. What should I do differently?
Perhaps you talk to your supervisors as a consultant to his clients. For example you would say “I have done the calculations. The results confirm your expectation. You do not need to check this. You can trust me.” Instead you need to talk to them as a scientist to a scientist. A scientist would say “This is how I did the calculations. This is the input and this is what comes out. Someone can redo the calculations. If I overlooked something I would like to know.” In a university we communicate the scientific way.
I am available to supervise this project. Do you agree?
Supervising a graduation project is not a good idea when you have graduated yourself recently. Surely you would be able to coach the student very well but the problem is the grading. Your brain will compare the grade you need to give to the grade your project received. Your brain will not allow a grade as high as yours. The psychology might be that your brain confuses your past skills with your present skills. Anyway, this upper bound would not be fair to the student.
My supervisor gave me a list of comments on my report. What should I do with these?
This is a peer review situation. Many authors think that reviewers ask questions because they want to know the answer or want to start a discussion. However, reviewers make comments hoping that these will improve the publication. A useful reaction to a review comment is either “Agree; the text has been adapted (see high light on page ..).” or “Disagree because ...”. So, take the list of comments and write your short reaction under each of these. At the same time improve your report. It is easy to disagree with most comments but this does not work. It would be ignoring the work that the reviewer did for you. So, do your best to interpret each comment in a way that you can agree with.
Why does nobody receive the grade 10 for his or her MSc project?
In theory, a 10 is given to 1 out of 100 students. Hardly any project supervisor has been involved in 100 or more projects. So they cannot know that your work is worth a 10.
You have your opinion and I have my opinion.
Sometimes I have a discussion with students. I do my best to formulate the argument clearly but often a student is not convinced. He or she thinks that it is a matter of opinion and everybody is entitled to his or her opinion. Of course I appreciate the independent thinking but science is not a matter of opinions. Please present your sound argument ... and please listen to mine.