becoming an “Intelligent Enterprise”
There is a plethora of published articles on the benefits to be realised if companies invest in a more connected digital asset base. Whether the operations are in generation or distribution infrastructure (or both), there are forecasts aplenty of enhanced reliability, efficiency, safety, and reduced costs. By becoming an “Intelligent Enterprise”, companies will be able to release these benefits. This means increasing data capture through the addition of more sensors and using this new data in an effective way to create a dynamic (real-time) link from assets via cloud-enabled platforms to decision-making processes. It has been reported by the US Federal Government that the US power generation sector will spend an estimated US$46 billion over the next 10 years upgrading generation assets and infrastructure. Similar spending is also estimated to be being planned in Europe with at least US$133.7 billion over the same period1. These investments cover the entire value chain, which will include the development of “smart grids”, will also allow power generation from renewables to be more effectively integrated into the supply/demand equation. This brings the added potential benefits of reduced environmental impacts through lower overall carbon dioxide emissions. This article will focus on the current developments in power generation plants to analyse the benefits and risks associated with adoption of a more digital asset base.
Like other industrial sectors, the commentary around digitisation is extensive and highly persuasive. However, there is little discussion around the potential risks that may arise out of this evolution. Through regular risk surveys over the past 5 to 10 years, Willis Towers Watson engineers have observed changes to the power station assets where the companies have invested in digital infrastructure. But have the benefits of a more efficient and safer operating environment started to emerge?
Figure 1 to the left shows power sector loss for the past 20 years which suggest the over this period losses in the sector have increased. This loss trend has occurred over a similar period where we have observed increased automation and analysis of the resultant data. Based on the claims regarding increasing digitisation, companies should have expected to see an improvement in profitability and safety. There would appear to be other elements influencing the performance in the sectors; in an attempt to better understand what has happened, a closer look at the loss data and what has changed within power plants is necessary. Figures 2-4 overleaf shows various charts of the total loss in US$ for different sub-categories of the power sector:
From the charts in Figures 2-4 to the left, it can be seen that the majority of losses have come from the conventional power generation sector. The loss data has been further analysed to reveal that most losses are derived from gas-fired power stations. The total infrastructure losses are significantly lower than the conventional losses but show a similar trend to the conventional sector losses. Renewables losses appear to be totally uncorrelated with the other two sectors and have a much smaller total loss level. Figure 5 overleaf shows the total loss in US$ for gas-fired power stations. This is a subset of the conventional power sector total losses displayed in Figure 2.
From an analysis of the loss data, it would suggest that a significant portion of power sector industry losses are occurring regularly on gas-fired power stations. As we have mentioned earlier, this does not appear to fit well with the postulate made by many articles in the field that state that increased automation results in a more efficiently run safe facility. Given this suggested contradiction, we will review some key factors around gas-fired power station operations to see if this sheds any light of the situation.
Two of the key areas that have seen significant increase in data capture and analysis are gas turbines and power transformers. These are examples of where predictive maintenance is becoming the standard operational practice that can impact both production and maintenance departments. Gas turbines Over the past two decades, manufactures have steadily increased the instrumentation and control equipment on their machines, which has significantly increased the telemetry amount that has been collected.
A typical power plant will produce two terra-bytes of data per day, which is highly valuable for tweaking general performance and predicting possible issues that are slowly developing within the gas turbine power train. The major gas turbine suppliers have been monitoring the operational data for the last two decades for clients who have purchased the monitoring option with the service agreement with the OEM.
OEMs would batch-process the information using a rules-based deviation alarm system to identify performance divergent from normal operations. For example, a deviation of compressor vane temperature might indicate fouling of the compressor or misalignment of the compressor section of the turbine. Historically, the site staff would tend to spot the deviation and ask the OEM to review the data. The OEM would advise a solution that would allow a running regime until an outage where the issue could be corrected. This monitoring process, while often effective at preventing major losses, would often require an outage at short notice, which would be expensive to organise; furthermore, the loss of production may occur during a lucrative operational period. The OEMs have now developed relationships with software companies to provide an earlier prediction of possible gas turbine issues. Algorithms are now used by advanced systems which calculate the rules base and modify it to accommodate load fluctuations and ambient temperature changes. This allows for a more refined rules base which can include rate of change as an alarm feature, which can be recognised by the algorithm system and not be noticeable as a trend by operator visual inspections.
For example, should there be a 5°C shift in temperature in a gas turbine burner, it could take over two months for a trained operator to notice the change. In contrast, an algorithm-based system would identify the trends and alert the operating company much quicker. The other area of improvement is the ability to remotely tune a machine. In the past, pressure monitoring connections had to be made locally and the engineer and analyser had to fly to the site, which was restrictive and expensive. Now with the latest digitalisation software, switches can be operated by the site engineers to allow temporary remote access to the controllers to allow the tuning of the combustion process to take place. This results in a quicker solution for the client and a more timely intervention by the OEM to prevent long term damage occurring due to non-ideal combustion. These two examples are relatively new additions to the tasks that remote OEM monitoring centres can undertake, over and above their more common tasks such as vibrations monitoring. Power transformers Transformer condition monitoring has been traditionally carried out by discrete testing or online dissolved gas analysis of the insulating oil. The trend over the last five years, with improved connectivity and digitisation, is the installation of complete transformer monitoring and supervision packages. The package available now includes high voltage bushing monitoring, which uses a capacitance tap on the bottom of the bushing; this allows the capacitance to be monitored online along with potential partial discharge activity. This in turn provides a real-time health indication of the bushing condition, which is improved when load current is measured and added to the supervision process. The digitisation process is considered complete when temperature monitoring, traditionally monitored with manual records, is added to the supervision system. This then allows algorithms in the transformer monitoring system to provide warnings if the transformer is being over loaded, cooling system issues, and a condition monitoring system that will compare phases of the transformer to each other. The benefit to the asset manager is that long-term slow changing temperature and dissolved gas trends can be identified earlier before becoming noticeable to operators. This maximises the potential for investigation of the issue and correction before serious damage can occur.
There are other factors that influence power station operations and the potential for a loss event to occur. The following list is not intended to be exhaustive, but highlights the main topics to consider:
From this list of potential other factors that could well be contributing to losses in the sector, at the present time it is clear that increasing the digitisation of operating assets is not going to cover all these areas. Therefore, in analysing the current loss record and the challenges faced by operators, we would suggest that the benefits being claimed for adopting a greater digitisation strategy are slightly premature. Directionally, they are probably correct but either operators need to eliminate these other factors or the digitisation infrastructure, in some way, needs to be able to measure these factors.
Roger Hughes is Senior Engineer, Natural Resources, Willis Towers Watson London. Roger.Hughes@WillisTowersWatson.com
1 https://www.powermag.com/how-digital-intelligence-can-be-a-difference-maker-for-power-plants/ and https://fas.org/sgp/crs/misc/R45156.pdf
