Automating a Crane with Antisway

Antisway: don't try automating a crane without it. Here are some stories and advice from 22+ years of developing and implementing antisway automation for cranes. We began this development around 1990 with support from the National Science Foundation. Our original grant was to track ship motion at pier side to support automatic positioning for a ship to shore crane. When we started looking at the problem we learned that the real unsolved problem was eliminating sway so that loads could be lowered safely, so we switched horses.

My colleague Rob Overton then solved the mathematics and we demonstrated the first computer antisway at Virginia International Terminals. After that we delivered a number of successful systems using various computing platforms and computer languages, all of that involved hard work but was never boring. The bottom line is, if a crane needs to be automated for any reason, the first job is antisway.

How does antisway work? We need to understand what causes sway in the first place. To move a trolley or bridge you've got to apply horizontal acceleration. But the hook or load is suspended by cables, so in order to transfer that acceleration to the load, the cables must rotate from vertical. That is sway and therefore unavoidable if you want to move the load. Therefore there's no such thing then as avoiding all sway but the important thing is to have little or no sway left over once the move has stopped.

There's some additional benefit if we can eliminate swinging once the trolley gets to full speed. The reason here is that accelerations from a heavy load swinging back and forth relative to the trolley (whether the trolley is moving or stationary) stresses all the structural components. So moving as gently as possible will extend the life of your crane's skeleton, cabling, and equipment. Longer life for ropes and bearings, at the least.

I've seen a number of mechanical systems designed to move loads rigidly, with complex cabling and supplemental motors. All the ones I've seen had been disabled or removed entirely, because of stress failure or system complexity, or both. It's never a good idea to fight Newton's law.

So, how do we avoid unwanted sway? Remember that the swinging action is harmonic motion, the period of which can be estimated from the length of a pendulum from the drum to the load center of gravity. In many cases, especially if the cables are not all hanging straight, the period can be hard to calculate and must be determined by swing timing trials.

Once we know the pendulum period, then it's possible to time accelerations such that when there's no acceleration there's no sway and the ropes are vertical and stable, whether the trolley is moving or stopped. This technique is well known and is called "bang-bang," because it means that all accelerations are applied in pairs. The first acceleration generates some sway and the second one, some time later, cancels the sway created by the first one.

Take some string and a suspended weight and move your hand a short distance and stop, causing the weight to swing. Now try to make a second motion, timing it so the weight is not swinging when you are done. A few practice motions and you'll get it right. Professional crane operators take some serious training to do this on a real crane, and years of practice to do it really well.

A further complication is when the load is being hoisted up or down in the process, which presents some serious timing problems. It's possible for a computer or PLC to control the accelerations to always result in a stable load no matter what. Those who have succeeded at this always have a "secret sauce" program that produces the desired result.

The simplest antisway implementation is to read operator demands (joystick or speed button) and apply acceleration patterns to do the job. If the programming is done right, the result is always the same: no swing once the trolley stops. It's always fun to challenge an operator to "make it swing" with antisway on, and observe that they really can't.

From an operational perspective, applying antisway to operator demand isn't always a popular approach. Operators with joystick control are used to applying their own acceleration patterns to accomplish the same thing. They learn their own patterns and get frustrated when antisway is turned on and the crane does not do exactly what they command. I've had seasoned operators complain loudly and stomp out of the cab for this very reason, while newer operators are elated that they can just punch full speed then full stop and get a nice stable move.

I believe the real payoff for crane antisway is the ability to push a button and produce automatic motion to a prescribed destination. Of course this requires instrumentation so that the trolley (and bridge) position is known in real time. Here a separate automatic move process decides directions and speeds to send to the antisway process, which in turn sends speed demands to the motor drives. The only hard part is deciding when to command zero speed at the end, so that the antisway process will cover the exact distance to arrive with precision at the destination.

A benefit of doing automatic move with antisway is that the load never swings beyond the trolley at the end of a move. That is, unless you want the load to swing out (for dumping bulk). In that case, instead of commanding zero speed at the end, you command full reverse. The trolley still stops momentarily at the destination point but the load swings beyond. I've had to tell a customer there was no extra charge for this "feature."

Besides the instrumentation, automatic move presents additional constraints. Primarily, the actual speed in ft/sec or m/min must be a a true function of the signals to the drives. This is not a constraint for antisway because antisway only cares about time, acceleration, and period. The automatic move on the other hand must know exactly how fast the crane will move in order to decide the speed and to predict the deceleration distance.

Another issue with automatic move is the accuracy of position measurement. I've had a situation with 250 meter/minute trolley motion and 8Hz position update rate. I had to point out to the crane engineers that at full speed the trolley moves more than a half meter in 1/8 second, so asking for 10mm stopping precision was not reasonable. I've also had a crane where a single turn encoder was used to measure distance along a 50 meter beam. The roller shaft was fitted with a reduction gear, making the distance increment nearly 400mm. Same stopping problem there. Some careful prediction algorithm solved both of these problems.

Once an adequate instrumentation system is installed, now one can envision a very complex motion menu. With two dimensions of motion such as an overhead or gantry crane, we can do almost anything automatically. Note that if the two directions are independent, one direction of motion will usually finish first, due to the difference in maximum speeds. We've done steering to slow down the faster direction allowing for true straight line motion where needed.

We have also implemented maneuvering, where a path to a destination needed to be segmented into several intermediate destinations to get around fixed obstacles. Even one case with a 100-ton ladle crane where certain obstacles needed hoisting to clear the load. In this case, new software calculated the shortest route on the fly, because there were too many possible routes to plot them out in advance.

Antisway is essential. We've helped out a small automation shop where they had been trying for months to automate a single crane for a customer, without antisway and of course unsuccessfully. We supplied some PLC code and made a one day visit and got everything perfect. The lesson is, to automate a crane you need antisway. Without it you will fail.