Are We Safer, Yet?

Safety has been in the news a lot lately.  Remember the Upper Big Branch coal mine explosion in April and the HUGE ongoing Gulf of Mexico oil debacle?   As always there is enough finger pointing going on to implicate a whole host of manufacturers and operators.  If we set aside all this media attention for a moment and get down to the nuts and bolts of safety there are really three main players:  Original Equipment Manufacturers (OEM’s), Operators, and Regulatory Agencies.

As a manufacturer of sensors we supply to the Original Equipment Manufacturers and they rely on us to provide enough technical information and support to safely build our products into their equipment.  Operators (and sometimes the OEM’s are the operators) rely on the OEM’s to give them enough information to safely operate and maintain the equipment.  Meanwhile the regulatory agencies let everyone in the chain know what are the safe limits for design and operation of equipment intended to be used in the particular environment.

Given the nature of operating in dangerous environments, even with all the players doing their job well – accidents are inevitable.  There is no doubt that the loss of life is tragic and devastating.  Coupled with all the peripheral losses (environmental, loss of income, hours taken in clean-up, etc.) the cost is largely incalculable.  But where do we draw the line between reasonable expectation of the inevitable and cautiousness to the point of disabling whole industries?  Furthermore, does more regulation even create safer conditions with the complexities of modern day operations?  Given that most accidents eventually come down to operator error should regulations be directed toward manufacturers or operators?

Here’s a case in point.  Today’s technology could be used to create a virtually accident-proof car.  Front and rear radar could ensure safe separation of cars in traffic, governor control could prevent speeding (and even slow things down in poor conditions), traction systems could prevent skids, alcohol breath analyzers could lock-out drunk drivers and so on.  This would be an expensive car, but it would save approximately 40,000 lives per year in the US alone.  It also takes a lot of the responsibility for safe operation of the vehicle off of the vehicle operator and puts it squarely on the manufacturer.  Would people even want to buy a “safe” car or would it infringe on their personal freedom of driving expression?  Would manufacturers even be willing to take on the liability for such a car?

The bottom line – can we ever be truly safe and at what point does the economic hardship of “perfect” safety outweigh the risks of inherently risky operations?

Getting Smarter?

Welcome to the BEI Sensors blog where we’ll be exploring a lot of interesting and perhaps controversial topics in the fields of sensors, measurement and control.  Issues that we face everyday and that shape the designs that we produce. Click the RSS button at the top right to join the feed and invite your colleagues to join in the monthly discussions.

This month’s topic is Smart Sensors which has been a point of discussion as long as I can remember.  What exactly constitutes a smart sensor depends on who you talk to and the sensor type.  When I worked with inertial sensors, smart meant loading all of the temperature compensation and offset curves into the sensor and providing a corrected output.  This was easier and “smarter” than trying to build a sensor with no temperature sensitivity and zero bias to begin with.  That was one level of  smart, but probably not one that works for all sensors.

Here’s another example:  for some sensors, being smart means developing a sort of watchdog mechanism that monitors the vital signs and can alert the operator in the event of either a hard failure, or for the really smart versions a parametric failure.  This could be especially useful for safety critical applications.  Of course this system would have to be smart enough to avoid “false alarm” indications.  A pressure sensor might be a good example where this could apply.

In some instances a few LED indicators show the health of the sensor and that is all that is required.  In it’s simplest form this could be a single “Power on” LED.  But this this really smart anymore?

With these examples can we put together a simple definition for “smart sensor”?  Here’s one thought: “A sensor can be considered to be “smart” when it provides additional information to the operator beyond just the parameter being sensed.”  That definition fits the examples provided, but would that be considered smart based on your experiences?

So it begs the question.  How much functionality needs to be added before a sensor is truly “smart” in your book?  What about sensors that have the primary sensed output plus an additional output, like temperature. Is that “smarter” than just the primary sensor? If you need to specify a sensor, how do you attach a value to “smartness” in a sensor and when does “smart” become “bells and whistles” with no added value?

If you’ve got an opinion or experience on this subject, then post a comment below and stay tuned to see what others have to say.

Engineering, Politics and Finance

I was always taught that you didn’t discuss politics in polite society since it was sure to result in some heated discussions and potentially ruined relationships.  Speaking on behalf of the engineering community though it seems we could use a little more critical thinking and a lot less politicking when it comes to governance of the financial markets.

Let’s take a look at the recent financial meltdown that started in the U.S. and take it down to its simplest elements.  Easy money (low interest rates and relaxed lending practices) led to banks lending money to people who were a really poor credit risk.  High risk ventures are by their very nature volatile since they are highly leveraged.  At some point the risk gets so great and the substance behind the risk is so tenuous that the whole financial system that has been built around they flimsy assumptions collapses.  Just like a bubble – hence the name.

This is not a brand new phenomenon.  Dutch flower bulb traders in the 1500’s were a prime example, but there are even more recent ones to look at:  Enron energy market trading and the dot-com implosion come to mind.  Any control engineer that has a lick of sense can tell you that when a system runs open loop (is unregulated or loosely regulated like the lending markets were) and a high gain (lots of money pouring in) it can easily get out of control.  What we need is an engineered solution to these kinds of problems since they are really system level problems.  Throwing politicians at the problem is sort of like hiring Martha Stewart to design a car.  It may look good, but it won’t run worth a darn.

It’s time to “engineer” a solution to prevent future bubbles.  I would like to propose that a simple PID control should do the trick.  Let’s take the difference between the money being poured into a market and the market value as determined by sales as the “market error”.  Apply a proportional control so the bigger the error, the stronger the feedback to reduce the investment.  Now we also need an integral control to force the investment to drive toward a specific market value.  Since integral control drives the error to zero over time, we can use time as a variable and further tighten the lending restrictions the longer the condition exists.  Lastly, the differential control portion minimizes the overshoot of the control and helps it settle down more quickly based on the rate of change so we can look at the rate of increase of investment as another element to dampen the enthusiasm for continued investment by limiting the amount of “speculative” investment (for the housing bubble example, we could require an owner to actually hold the house or live in it for a period of time in order to realize a gain from the sale or pay a huge tax penalty for example.

There you have it.  People respond to financial incentives, let’s just start tying the financial incentives in a logical way to the desired result and keep the politicking where it belongs – wherever that is.  I welcome your thoughts on the workability of this proposal.  Let’s put some logic and control into the political system.

Tagged with:
 

Feedback Anyone?

We’ve all had the experience one time or another of a less-than-perfect encounter with a product, person, or service.  Remember that store clerk who just couldn’t be bothered to come over to the register and ring you up?  How about the waiter who was less than enthusiastic about sending your undercooked meal back to the chef?  Or that recent backyard grill that you had to assemble without the critical special wrench that was supposed to be included?

You probably were a little frustrated, but really couldn’t (or didn’t) do anything about it.  Well, maybe you stiffed the waiter at least – but even that isn’t a very satisfying result.  You would have preferred to just have good service to begin with.  What you really want to have happen is to be able to give immediate, honest feedback and have it acted on, like right now, please!

Things are more straightforward in the world of industrial control.  Sensors become the eyes, ears and the nerves of a process constantly communicating back to a controller with an honest assessment of performance.  Temperatures that are too high get cooled, operations that are too slow get speeded up and processes that need to be synchronized, really are.  What could be simpler?

So I got to thinking, can we really apply the feedback rules of control systems to the living breathing world of humans?  I thought I might give it a try.  My first chance came when I sat down in front of a chatterbox at a recent theatre event.  Partway through the first few minutes of the performance I turned around and said, “Excuse me, but I was wondering if you might hold your voice down a bit while the musicians are playing?”  I smiled politely and then turned back around in my chair waiting for the huffy rejoinder.  None came and all was still.  At intermission as I stood up, the woman looked at me and said with a grin “That was the nicest way anyone has ever told me to shut up”  Score one for feedback.

I had a recent opportunity at work when the IT department had to reinstate my access to certain files because an “automatic” file manager had reset the access codes.  To be honest they did solve the immediate problem, however, just like a temperature controller that was still reading a wrong temperature, I suggested that they go the next step and actually set the attributes of the code so it didn’t happen again saving them (and me) a future repeat of the same email exchange. “Great idea!” came back the enthusiastic response.  Another triumph for feedback.

Now I’m emboldened to find other opportunities to try this feedback scheme.  I figure if I’m direct, polite and honest, I’ve got a chance to succeed.  At least I’m two for two so far.

 

The Wonderful World of Wireless

This has been a hot topic off and on over the years and with so many companies claiming wireless-ness in their sensors it’s a good time to take the pulse of the market and get some real world feedback on wireless adoptions. A quick review of some industry surveys indicates that the primary reason designers choose to go wireless is (drum roll please) no wires! Yes those pesky electrical conduits not only get in the way, but they actually cost something to install and maintain.

The laws of “no such thing as a free lunch” require that there is a tradeoff and that’s where your personal experiences come in. Naturally it’s hard to beat actual hard wires for bandwidth and carrying signals over a distance. By contrast most wireless systems have some sort of data latency effectively limiting their bandwidth and finding single sensors to transmit, say 500 feet is darn near impossible. These wireless sensors are basically radios after all and radio frequency interference can get to be a big problem when you have a lot of electrical equipment around in your factory.

Mesh networks are sort of a way around the long distance problem. By having a bunch of smaller “signal repeaters” scattered around, you can eventually get the signal from point A to point B, but that doesn’t help the bandwidth problem and now instead of a cable you have a bunch of “mini radios” scattered around the plant – each of which requires its own power source and a periodic battery change. So is this any better than the original wired solution?

My question to you is, with all of these manufacturers making wireless sensors available, where does their use really make sense? Here’s a chance to share your comments and experiences of successful (or unsuccessful for that matter) wireless installations that you’ve been involved in; what worked and what didn’t? Add your comments, below.

Tagged with:
 
<?php //akismet_counter(); ?>