Ampair 6000 progress update (II)

It has been some time since we updated everybody about our progress with the Ampair 6000, in fact since the 23 October 2009. Because of this I may get some of a rather complicated history out of sequence.

At the time we were waiting to switch on one Scottish site where they had not completed the local wiring; we were having major difficulties with a site in eastern England; and a site further south was nearly ready to commission.

The site further south was in fact in northern France. We switched on there and the inverter would not connect to grid.

We continued to investigate the eastern England site and it turned out that a major underlying issue was the inverter. We found it very difficult to get technical support on the issue we were seeing and we were beginning to wonder if it was not the inverter at all, in fact we believed we were chasing several different issues simultaneously. Eventually after almost two months we were able to find a work-around for the inverter. The people we were buying these from then announced that they had parted company with the manufacturer of the inverters. The manufacturer tells us that the problem we are experiencing is not their fault, we beg to differ and think we know what is their issue, but we have found a work around that is adequate. Unfortunately all the messing around with an uncontrolled turbine at this site has rather messed up other things which we puzzled away at and as we did so we made changes to other production units to test out the various theories.
Once we found the work around for eastern England we commissioned the French site which immediately worked except that it exhibited an odd rattle at some wind speeds. After thinking through the options, and having had further experience in eastern England, we put this down to a loose magnet and pencilled in a turbine head change to investigate. We had seen a loose magnet on one of the turbines and the symptoms were the same.

As an aside all the remote monitoring packages have worked fine at our own two test locations but not at any of the client sites except for briefly at the eastern England site during a period when unfortunately the sensors were inoperative. We continue to chase our tails on this which is a nuisance as we would really appreciate that data, but fortunately it is not something we included in the ‘sales’ package. At the moment it is a ‘freebie’. We continued to puzzle away at the east of England site which seemed to be seeing unique problems and experiencing terrible bad luck with a run of simultaneous faults.

We installed over on the west coast of Ireland which went smoothly except for the remote monitoring package. Apart from some absolutely awful weather that is – I was trying to teach the local installers in horizontal rain with floods lapping towards our feet.

Then we almost simultaneously commissioned the first Scottish site, a second Scottish site, and a second French site. The second French site was in a desperate hurry to have a unit because of some publicity deadlines and delivery problems from other turbine suppliers (there is a rather depressing history in France with small wind turbines over the last few years) so we let them have our own test prototype Ampair 3500 and explained that we would have to return and upgrade them to an Ampair 6000 at a later date. We were told it was working very well immediately after installation – we gave them all the other kit as per the Ampair 6000 to make the eventual change easier but we limited the power curve so as not to burn anything out.

All these were done with hand made circuit boards and if you count you will find that there are about six units in all which cleaned out our stock of hand made circuit boards. During this period we transited from one set of interconnect unit designs to another as a result of learning in the east of England site, so now we have a mixed fleet out there that we need to revisit. We also had cleaned out our first run of mechanical hardware and did not even have a test unit of our own left.

Whilst all this was going on we were bringing about thirty sets of mechanical hardware through our production system. At a certain point we realised that we were almost two months late with the castings. It took a while to get to the bottom of why this was but finally the foundry explained that the batch of castings had failed the hardness test post heat treatment. They had investigated this and traced it to a supply of ingots with a forged certificate of conformity. So they scrapped the batch of castings and started again. They also invested in an analyser for all their raw materials – they are a very quality conscious foundry with some very high profile customers so were rather relieved that it had happened to our product rather than one of their other clients, and don’t want it to happen again. This set us back about two months on the thirty or so mechanicals.

This in turn was causing knock-on problems elsewhere because we have a rather large order backlog, plus we wanted to completely reinstall at the eastern England site where we were rather suspecting the continual abuse (due to our run of bad luck) had caused a magnet to come loose and jam the entire system. This was after we went there one day to tune the system, observed 6kW output, and then it jammed up. Most unfortunate but at least a safe state to be in.

So by the Christmas period the good news was that blade production was just about getting ahead of demand, and the bad news was we were waiting on a batch of production electronics, and a batch of production mechanicals. The electronics kept on going right as the circuit layouts slowed down. We tried to assist a French client by doing a part delivery of some blades and some mechanicals, and to do this we worked through the holidays and postponed a factory move (we are trying to get into larger premises). They would have to wait for us to deliver the electronics and the remainder of the mechanicals at a later date, same as with eastern England.

Driving back from France in the snow on new year’s eve the phone rang to tell me that the Irish client had woken up in the morning to find his turbine had shed its blades. We immediately contacted all of the running turbines to shut them down. During January we analysed the failure which we traced to a change in the details of the manufacturing process that caused a fatigue failure. We decided to not only revert to our initial manufacturing process (remembering that we had been running a turbine for about nine months before the first unit went to a client) but also to massively reinforce the internal structure of the blade. We filed a HSE incident report with the BWEA for their new small wind safety (HSE) database and we briefed our colleagues/competitors about our problems at the BWEA technical committee meeting. There are enough problems in this industry without trying to sweep safety issues under the carpet and we are happy to share learning in this context. Needless to say this was a pretty devastating episode for us as we thought we had just about managed to juggle all the eggs and get through to settled production.

This set us back in two ways. Firstly all the blades in service – and the five sets that were ready to go into service – became instant scrap. Secondly there was an additional process cost and doubling of processing time to make the reinforced blades. So we slowly made more blades (with some learning and some additional scrappage as a result) and issued blades to two sites – three sets out because one set got stranded en route. These were to sites where the electronics are known to be OK. Unfortunately the Irish site must wait until new electronics are ready as the turbine head needs replacing as a precaution.

Right now we are working our way through the new production PCBs. There are four in a set so there are 120 PCBs in all. Eventually we got the PCB drawings finished, and finally sourced the components some of which had gone out of stock in the interim. We have produced one set of four to proof test them before we go into full production and three have tested fine. The fourth is not behaving and we have been through two build cycles trying to debug it. It has improved on each cycle but is still not quite there and later this week we are going to do a comparative analysis of our lab unit with the production unit, using the production contractor’s test equipment. These are fairly large surface mount boards that are manufactured in a clean room environment so one must work slowly, patiently, and carefully during the debug. As you can imagine this is not filling all of our clients with delight and they definitely do not all appreciate the “how long is a piece of string” nature of debugging.

To speed up the blade manufacturing process we also took a decision – once we were happy with the reinforced blade production and performance – that we would commission additional blade moulds. We have enough of these arriving in about a fortnight to treble the production rate versus where we were, or increase it sixfold versus where we are. That will be a relief as we have more catching up to do both with installed units and with the order backlog.

That brings us to the tricky issue of the order backlog. The good news is that we don’t ordinarily accept deposits and prepayments (we have accepted one for special reasons). The bad news is that on each occasion we have accepted an order in good faith then a few days later something else has set us back. Each time it has seemed to us that the responsible thing is to slow down and figure it out, but of course people who have placed orders don’t see it quite the same way. The still want us to keep to the predicted delivery dates we gave them. Actually not all think like that – some are more understanding, and some are less understanding. Some have now got masts installed or delivered or foundations in the ground, and some don’t. We will be able to give you all dates when we have that fourth PCB through its test. Until then it is always a few weeks away because that is how long it takes to test/produce/assemble/deliver and what we don’t know is when the test cycle will be adequately completed.

In the meantime we try to juggle everybody’s needs as much as we can and we are keeping our heads down and not seeking additional orders. I could try and explain what all this has done to the factory move or our cash flow but that is a story for another day. Likewise this has not been doing wonders for our steady-state business to have all this going on or a frozen factory move. As you will appreciate this is just the bare bones of an even more involved situation so please excuse me if I have missed out details.

Ampair 6000 progress update

A quick update on where we are with the Ampair 6000. Basically we have started installing and as a result are finding a few snags that we had not flushed out in testing at our own site.

On one site in Scotland the client wiring was not ready and so that installation has not been commissioned yet. We've since had a problem at another site and have asked the Scoittish to hold fire a moment in case we want to make a change to anything - it is a very long way away in case we switch it on too soon.

At a site in eastern England we used a new alternator design that has had bench testing but not had live wind testing at our site. We pulled it forwards in the build sequence because that client wanted the new alternator which is smashing from a performance perspective as it includes a few lessons from testing of the first batch. We've had a lot of problems which are probably because of this decision. We suspect that the alternator is the cause of the problems but equally the site is showing issues with earth leakage. Or it could be a combination of both alternator-inverter and inverter-grid. To add insult to injury our nice remote monitoring system has decided to be uncooperative on this site of all sites. Need less to say we are furiously trying to get to the bottom of this as it affects all turbines that naturally fit into the production sequence after that one. And unfortunately each time we go to site the wind does not blow much. We've struggled with this for two weeks now.

Another site further south has been installed and is going live any day now with the initial batch of alternators which will help us in trying to flush out these problems. We are also building various test rigs and looking to fly a new style alternator here at the factory if the wind cooperates (which it did not when we last tried).

The assembly sequence is going fine and the mast installs are going fine so that side of things is good. The remote monitoring is still a work in progress but is making progress. We are trying to improve a few installability issues with our electrical package - to be truthful more maintainability than installability. These have become more evident as a result of trying to troubleshoot the new alternator unit.

So not quite where we'd like to be, and we are very much scratching our heads on the new alternator unit. But progress none the less.

As it happens two days ago I was at the BWEA annual conference in Liverpool and a colleague at another small wind turbine company told me they'd recently had to switch off all their machines due to a manufacturing defect they found. He said he was getting a lot of grief over this and that clients are absolutely expecting right first time. He found that rather unfair as even Boeing are late by two years on their 787, and often major auto manufacturers are late with new product development or have to do recalls. And these major companies have enormous resources which for sure small wind companies seldom do. His view is that consumers have become detached from the underlying engineering reality as a result of virtual reality and marketing. Anyway I guess we'd better plug away at learning how to solve our issues and this blog post is just to update everybody the same way I updated the visitors to our stand at the BWEA conference.