Automation of a giant hydrodynamic lab

Ships, rigs and other vessels depend on research for getting better and more efficient maritime transportation means. Hydrodinamic calibrators - like TPN in Brazil - play important place in this role and need automation to work.

Procurando pela versão em português? Clique aqui.

Research in the naval field of engineering depends on large resources such as supercomputers for simulations - called numerical simulations - to simulate the sea or river behavior against a vessel built by naval engineers. Besides the computer simulation, scale models construction and test in a water tank of big proportions and able to generate waves in the water is an important step to compare and calibrate engineering parameters that cannot be perfectly simulated in computers. Brazil has some labs to perform such as this research. Depicted above, the main entrance of TPN - Tanque de Provas Numérico or, Numerical Offshore Tank in English - inaugurated in 2009 can be seen, one of those labs.

I must say as the starting step of this article that I am not a specialist in naval engineering or in hydrodynamics. The reason why I'm writing this article is that I feel glad I had the opportunity to contribute, on the automation side, to the construction of the Hydrodynamic Calibrator (HC), one of the instruments installed inside the TPN - Tanque de Provas Numérico or, Numerical Offshore Tank in English - and I'm happy to share this moment with you, reader. The year? 2007. That's the year I began to be involved in the HC project. The University of São Paulo (USP) was in charge of building the TPN lab (Numerical Offshore Tank lab) inside their facility with funding and support from the government and Petrobras. USP already had a portion of the lab, comprising only an old supercomputer and the challenge was to move to a new lab and build the structure for a new supercomputer, classrooms, a hydrodynamic calibrator (HC), and other resources needed. A hydrodynamic calibrator is a big tank, filled with water, for simulating scale models of vessels, placed in or on the tank. The objective is the generation of waves in the water, according to a certain pattern that depends on the geographic region of the sea and others, to check what is the behavior of the model in such wave condition. Sensors are placed in the scale model to evaluate mechanical resistance, movement, etc. that are later analyzed by naval engineers.

The Hydrodynamic Calibrator (HC) of TPN, after finished

The TPN's HC was under projecting and Prof. Pedro Cardozo de Mello was in charge of the HC project and construction, looking for automation equipment to move the flaps - big pieces of metal arranged along the tank edges - for wave generation. Salesmen from our company and distributor that were taking care of the sales side of the project came to me with a request: "We have products for this HC project, servos, CPU and network, etc but the USP requested us to calculate how much time it takes to a command sent from a computer to get to a specific servo to move a flap. We need your help to calculate it. We must achieve less than 10 milliseconds for this command to get to the servo". That was the exact moment I started getting involved in the HC project. We were competing against 2 or 3 other automation companies. That was an important project to us, in sales volume: more than 100 servo axes and a substantial amount of accessories and CPUs. One of our biggest distributors of the time was taking care of supplying those parts and they were putting pressure on me for getting the calculations done as soon as possible. At that point in time, I hadn't experienced before such a big project or such a big calculation. I started building in a paper all the hardware and its connections, looking at the manuals for the processing time calculation for each piece of equipment, network, etc. I remember to ask help to some technical/marketing people in the company that was responsible for our business in Brazil (it was the American branch at that time) and I confess I got disappointed with lack of help, even telling me that that project wouldn't work. Later, after the project was up and running, I got more disappointed and sad (not to say angry) with those "no-helping" people - that were supposed to help me - saying to me "It was easy, huh..." after I told them how I devised a way to solve calculations and a way to make it work, with help of other people (that didn't have the compromise to help me, such as Engineer James Smith that I respect a lot! God Bless him!). I'm trying to forget those episodes, unfortunately, it is not so easy to me... anyway, let's get back to interesting and happy parts! I could make all the calculations and, in my first try, as the solution quoted initially by salesmen, I got about 14 ms. After an internal discussion, we decided to go to new hardware, not launched yet in the western market but much faster than the one we were proposing to the customer, a way to solve the problem. I had contact and help from a manager (engineer) that is one of the engineers I respect the most for helping people and for its knowledge: Mr. Nobuyuki Matsuoka! Matsuoka-san helped me with some suggestions about how to make the software and after the calculations finished I got less than 10 ms (something like 8 or 9 ms). I was happy and apprehensive at the same time: besides I trusted in my calculations and my technical capabilities, it meant we would be approved technically and the challenge of making the software work would start. We sent the calculations to USP for the analysis and possible approval for purchasing our products. The first answer was that we were not approved, because of pricing: another competitor reached the performance expected by USP, with a better price. Ok, game over, then to us... To our surprise, 2 weeks after we got the negative answer from USP, USP came back to us telling us we were the second in line in their rank for suppliers and the first supplier approved initially was then discarded. That supplier had problems when answering technical enquires on the time calculation so, USP decided to move to purchase our solution.

HC during the construction phase

After we were approved and the purchase order was issued to us, works for developing the software started. I remember Prof. Pedro and I worked in a simulation and prototypes: first with motors with CPU, no-load, on the workbench. It worked! After that, we moved to test with 2 or 3 flaps with the tank still under construction. It was good again! Then I was happy and relieved. The year was 2008 already. The work proceeded, while the tank was being built. We gave training to technical people involved in the HC project at USP, Prof. Pedro worked on expanding the prototype to a full-scale software, for all the 148 flaps to be placed along the tank edges. In that phase, I was visiting USP twice a week to support USP in developing the software and to solve other technical issues.

Motors and electrical arrangement for moving the flaps

After the installation of all flaps, some other technical issues such as adjusting the control loop gain parameters due to resonance with the mechanical system happened but could be solved satisfactorily, after some struggling. Finally, in 2009, the complete lab - including the HC - was officially inaugurated. Hidden figures such as me were not in the official event... I don't care for such formalities. What I know is that I was happy to contribute with such a special project! You, reader, may be curious about the results... I have a surprise to you: watch the video I made below! It shows tests done at the finished tank as proof of concept. Interference of waves, a concentration of energy and, even drawing in the water can be seen in the video. You will find also many videos from Prof. (now, Dr.) Pedro on YouTube!

As always, I hope you enjoyed this history and feel free to send feedbacks!

Best regards! Thiago