Making great ribs with data analytics

I love to cook.  Italian food, Japanese food, any food.  You name it, and I have probably tried to make it at least once.  But one of my favorite things to cook are good ole’ ribs.  For the last few years, I’ve been trying to perfect smoking ribs on a gas grill with varying degrees of success.

My setup is fairly complex.  I have a Nexgrill infared grill.  On the left side, I use loaf pans to heat up soaked cherry wood chips to create the smoker.  After heating charcoal briquettes in the loaf pan and letting them get to temp, I put the wood chips on top.  They begin to smoke almost immediately.  With a little tin foil (holes poked to allow smoke to puff away), I have a decent makeshift smoker.  This isn’t a big green egg or offset smoker, so I can’t expect to sell these at the county fair.  But it’s pretty darn good.  I use the water from soaking the chips for moisture and put all that in the grill.  The ribs go on the opposite side.  I smoke for 4 hours, then wrap the ribs and cook another 2 hours, and then finish for 1 hour at a higher temp.  The last 20 minutes is where the BBQ sauce gets added.

This has worked pretty well.  But I never seem to cook the ribs enough.  They are ok, but they’ve always had a bit too much chew.   I couldn’t understand it.  I tried cooking them longer, wrapping them, adding liquid … I even tried to boil them once.  Don’t tell anyone.  The temp gauge on the grill would always be between 250-275.  That should be plenty high enough for smoking.  Some would say too high.  And then it occurred to me that I was measuring at the top of the grill and not where the actual food was.  So I placed temp thermometers in the grill near the grates.  This is when my problem was exposed.

I turned up the temperature and let the temp on the grill get to ~430F.  Way too high for smoking ribs.

But using an iDevices iGrill thermometer, I attained a different reading near the grates:


There is a 200+F temperature difference between the temp on the grill and the temp at the grates.  For those who are wondering, the iDevices was calibrated when I first received it.  I’m pretty sure it’s accurate – definitely way more accurate than the temp on the grill.  And I get similar results using various thermocouples.  So what’s happening here.

I have determined through other measurements that a very simple phenomenon is causing my ribs to be undercooked – heat rises.  A lack of air circulation inside the grill is resulting in the hot air to stagnate at the top of the grill and never make it to the surface of the ribs which is why they never seem to cook enough.  When I place the iDevices thermo near the grill thermometer, it rises dramatically and comes within 25F of the grill thermometer.  So it isn’t that the grill thermometer is cheap or inaccurate.  It’s fine – it’s the location of the thermometer that’s the measurement system problem which was confusing me as to what exactly the problem was.  And therein lies the crux of the problem – I was attempting to measure a system at a place other than where the value is being added.

OK.  Let’s be honest.  This isn’t really data analytics so much as measurements and analysis.  But I would argue that there really isn’t that much of a difference.  The concepts of data analytics are worthless if the data itself is completely flawed.  And in the end, you’re taking measurements and making decisions.  Without these measurements, I’m guessing at both my problem and potential solution.  I need to use the scientific method to solve this: come up with a hypothesis, determine a measurement plan, measure, analyze results, determine conclusion to hypothesis.  My hypothesis was that the temperature at the grates was different than the temperature being measured by the grill.  By using a better placed measurement system, I am able to better adjust my system parameters to achieve the desired result.  My PDCA cycle is far more effective because I have a much better understanding of the actual problem and the effectiveness of any solutions I’ve attempted to put in place.

Operations that attempt to adjust manufacturing setup/performance by using business profitability or global metrics (i.e., productivity) as their measurement are not measuring their system effectively.  You are taking a non-real time measurement that isn’t at the point of work and using that for decisions.  This going to be, at best, marginally effective.  By taking measurements geared around some kind of hypothesis at the point where the value is being added results in much better information.  I can determine where my problem areas might lie, adjust the system parameters more effectively, very quickly determine if those adjustments had any meaningful impact.  In Kaizen, Time Observations are so effective because they measure at the point where the value is being added.  With today’s technology, you can get time observation-like data all the time and have intelligent systems that tell you when something isn’t quite right or isn’t quite optimized.  The PDCA cycle is far more effective, once again.

Armed with better, real-time information, I made the adjustments needed to perfect my production system.  And, yes, the ribs were amazing this time.

*** on a side note, I do believe a good solution to this problem would be to place a small fan near the top of the grill and include another thermometer on the side near where the food is being cooked.  I’ll be trying this out.  If this works, I see a potential business opportunity in turning gas grills into effective smokers for the backyard grillmaster.


Spread the word. Share this post!

About the author

Greg McFalls is a recognized leader in the field of custom software solutions in lean operations environments. MCFALLSTECH was started to help as many companies as possible be more competitive in this age of advanced technology.

Leave A Reply

Your email address will not be published. Required fields are marked *

Contact us to discuss how we can help you grow your business.

We're here to help tackle some of your most challenging problems.