My entry to the PV industry twenty years ago came about because I was a quality technician and toolmaker in an automotive manufacturing facility. The automotive industry was facing tariff cuts, huge volatility and was a precarious place to work. After a chance discovery of a PV retailer seeking to implement a quality system, I jumped out of automotive and into PV under the naïve assumption that “PV must surely be on a massive trajectory and therefore be far more stable than this crazy roller coaster of an industry.”
Right on the trajectory, wrong on the stability!
Fast forward 25 years and like you I’m sure, I now find myself listening to endless claims from PV manufacturers all around the world about their quality systems, processes, materials and promises.
The vast majority are well intentioned and all are well aware that quality control is essential. There is an increasing desperation to differentiate by demonstrating endless compliance stamps and approvals. However, it struck me recently that I have heard no-one talk about the real science of quality.
All those years ago I was taught some golden rules about variability and quality in a complex manufacturing environment but one resonates most loudly and to be honest; it’s the one that worries me most about our industry.
I hear everyone talking about 100 percent inspections and in some cases 200 percent or even 300 percent inspection. “An absolute guarantee that we will catch any and all faults assuring perfect quality every time because we test every single one”. Wrong.
In a wonderful example of human nature and psychology, it was demonstrated to me that if you give ten people a simple task, then someone will get it wrong — despite their best intentions.
I was taught this and went on to teach it for many years and the results were always, always the same. Humans are imperfect and in reality, machines are imperfect too. This is why modern manufacturing facilities do not rely on 100 percent inspection if they are truly serious about quality.
The solution?
Let me start by saying, (perhaps stupidly) that I have no shares or pecuniary interest in quality systems, products or services.
What I learned is that Design of Experiments (DOE) and Statistical Process Control (SPC) were incredibly powerful predictive tools that could yield astounding quality results by getting to the crux of variation — before it crept into your processes. I saw it utterly transform the output quality of a manufacturing facility and the thinking of those operating the machinery; albeit with a fair bit of change management thrown in!
I’ll give you an example. One of the products we made was an air conditioning hose for a car. We rigorously tested and inspected them in all sorts of ways but ultimately, rejects slipped through. We undertook a highly detailed DOE experiment delving deeply into every step of the process, material variation and measurements, and found a hit list of processes that seemed to offer the best opportunity — by being the highest risk. DOE uses a systematic, mathematical approach to work through this.
More experiments were undertaken and we quickly realized the tool wear rate was pretty minimal and could be easily checked so we put a process in place to do that, but we still hadn’t discovered the true root cause. Through DOE we could “smell it” however; we knew we were homing in on the real issue.
Despite being a new machine of very high quality, one day while we were monitoring some records of variation over a 24 hour period, we noted that the variation changed according to the time of day. We quickly correlated this with temperature and ran more experiments. Before long we realized we had cracked it; beyond a given range of temperatures, the viscosity of the hydraulic fluid changed significantly enough that the valving could not cope and pressure variability rose dramatically to around five times the normal deviation. Although we were dealing in thousandths of an inch, it was highly, statistically significant.
I recall wild excitement as we pondered solutions. One of our engineers suddenly raced out and came back with a temperature probe and an oil cooler, which he’d pinched off his own car. Another called our oil supplier to get temperature and viscosity specifications. Within a few days, the variation was effectively gone; we implemented alarms for temperature, a regular replacement schedule, specification for the oils and stopped 100 percent inspection. Our reject rate virtually ceased.
I have not visited every PV factory in the world but I have visited many and worked in several. I know why some didn’t use DOE and SPC — the time and cost involved could not be justified against the volumes in years gone by. But it can now.
I am tremendously hopeful that there are many PV companies using DOE, SPC and other more modern variants, but I’m not hearing about it — all I’m hearing about is 100 percent inspection. It seems to me that there is a golden opportunity for those truly leading in quality to stand up and genuinely differentiate themselves.
100 percent inspection is not a 100 percent guarantee of quality — that’s a guarantee.
http://www.renewableenergyworld.com/rea/blog/post/2012/03/could-pv-manufacturers-who-are-genuine-about-quality-stand-up-please
Right on the trajectory, wrong on the stability!
Fast forward 25 years and like you I’m sure, I now find myself listening to endless claims from PV manufacturers all around the world about their quality systems, processes, materials and promises.
The vast majority are well intentioned and all are well aware that quality control is essential. There is an increasing desperation to differentiate by demonstrating endless compliance stamps and approvals. However, it struck me recently that I have heard no-one talk about the real science of quality.
All those years ago I was taught some golden rules about variability and quality in a complex manufacturing environment but one resonates most loudly and to be honest; it’s the one that worries me most about our industry.
I hear everyone talking about 100 percent inspections and in some cases 200 percent or even 300 percent inspection. “An absolute guarantee that we will catch any and all faults assuring perfect quality every time because we test every single one”. Wrong.
In a wonderful example of human nature and psychology, it was demonstrated to me that if you give ten people a simple task, then someone will get it wrong — despite their best intentions.
I was taught this and went on to teach it for many years and the results were always, always the same. Humans are imperfect and in reality, machines are imperfect too. This is why modern manufacturing facilities do not rely on 100 percent inspection if they are truly serious about quality.
The solution?
Let me start by saying, (perhaps stupidly) that I have no shares or pecuniary interest in quality systems, products or services.
What I learned is that Design of Experiments (DOE) and Statistical Process Control (SPC) were incredibly powerful predictive tools that could yield astounding quality results by getting to the crux of variation — before it crept into your processes. I saw it utterly transform the output quality of a manufacturing facility and the thinking of those operating the machinery; albeit with a fair bit of change management thrown in!
I’ll give you an example. One of the products we made was an air conditioning hose for a car. We rigorously tested and inspected them in all sorts of ways but ultimately, rejects slipped through. We undertook a highly detailed DOE experiment delving deeply into every step of the process, material variation and measurements, and found a hit list of processes that seemed to offer the best opportunity — by being the highest risk. DOE uses a systematic, mathematical approach to work through this.
More experiments were undertaken and we quickly realized the tool wear rate was pretty minimal and could be easily checked so we put a process in place to do that, but we still hadn’t discovered the true root cause. Through DOE we could “smell it” however; we knew we were homing in on the real issue.
Despite being a new machine of very high quality, one day while we were monitoring some records of variation over a 24 hour period, we noted that the variation changed according to the time of day. We quickly correlated this with temperature and ran more experiments. Before long we realized we had cracked it; beyond a given range of temperatures, the viscosity of the hydraulic fluid changed significantly enough that the valving could not cope and pressure variability rose dramatically to around five times the normal deviation. Although we were dealing in thousandths of an inch, it was highly, statistically significant.
I recall wild excitement as we pondered solutions. One of our engineers suddenly raced out and came back with a temperature probe and an oil cooler, which he’d pinched off his own car. Another called our oil supplier to get temperature and viscosity specifications. Within a few days, the variation was effectively gone; we implemented alarms for temperature, a regular replacement schedule, specification for the oils and stopped 100 percent inspection. Our reject rate virtually ceased.
I have not visited every PV factory in the world but I have visited many and worked in several. I know why some didn’t use DOE and SPC — the time and cost involved could not be justified against the volumes in years gone by. But it can now.
I am tremendously hopeful that there are many PV companies using DOE, SPC and other more modern variants, but I’m not hearing about it — all I’m hearing about is 100 percent inspection. It seems to me that there is a golden opportunity for those truly leading in quality to stand up and genuinely differentiate themselves.
100 percent inspection is not a 100 percent guarantee of quality — that’s a guarantee.
http://www.renewableenergyworld.com/rea/blog/post/2012/03/could-pv-manufacturers-who-are-genuine-about-quality-stand-up-please
No comments:
Post a Comment