Demand Logic analysis of building management system at King’s College London

Article from the December 2014 edition of the CIBSE Journal written by Liza Young.

A new data platform that extracts and analyses information from the existing BMS (building management system) has made annual savings of almost £400,000 at King’s College London. Liza Young finds out how Demand Logic works.

1 Introduction
2 Immediate benefits
3 Energy insanities at King’s
4 Closing the performance gap
5 An exciting future
6 Automation

[edit] Introduction

Finding energy-saving opportunities in heating, cooling and ventilation systems of large, complex buildings, can be like searching for a needle in a haystack. However, a new web-based platform that extracts and analyses data from building management systems (BMS) has identified 47 opportunities to save energy at King’s College London – which have, so far, resulted in annual savings of £390,000.

The system – developed by online energy efficiency company Demand Logic – uses ‘big data’ techniques to continuously stream and analyse thousands of BMS data points, including control signals, sensors, valve positions, set-points and meters. It stores this information in a cloud database, and uses it to provide a suite of infographics to help find common, but highly wasteful, faults – such as items of plant left in manual control, out-of-hours running, or simultaneous heating and cooling.

The primary aim is to find low-hanging fruit by identifying the ‘energy insanities’ that are still happening in commercial buildings. The extracted data has the potential to be used alongside dynamic simulation tools, such as EDSL Tas, and resource management and reporting software, such as Verco’s Carbon Desktop, to verify energy projections – and identify energy-saving opportunities – when designing new buildings.

[edit] Immediate benefits

Tom Randall, development manager at Demand Logic, says: ‘Many buildings – even from new – could be riddled with problems, partly because there are not many people onsite to look under the bonnet of what their BMS is doing. ‘It’s cost-effectively fast-tracking the snagging at the building optimisation phase, which can take two to three years – at the point where it’s most painful to the client, in cost terms.’

By avoiding unnecessary plant operation, the platform can help extend the life of equipment, and support condition-based maintenance. It allows facilities managers (FMs) to monitor equipment and focus maintenance – and to replace problem equipment, rather than playing a ‘big game of roulette’ by arbitrarily replacing working equipment at set intervals.

Another application of the system is during commissioning and the defects liability period. It can monitor all building services over extended periods, and ensure witnessing does not take place until services are within performance criteria. It can then monitor performance during defects liability, providing robust evidence for identifying issues and assigning responsibility, while avoiding ‘finger pointing’. The platform is aimed at a key group of people in the building – the ‘actors’ – who have access to the plant, and can act on the findings. ‘Onsite FMs are the best-placed professionals to pick up and run with this,’ says Randall.

[edit] Energy insanities at King’s

At King’s College, where the pilot project has been running since January 2013, the system has so far unearthed 47 energy-saving opportunities, 38 of which have already been addressed, saving an estimated 2,500 tonnes of carbon per year. The problems at the college, which spends £5m a year on energy, turned out to be diverse. Boilers were rapidly cycling on and off because they had inadequate load, and several conflicting temperature set points had been applied to the same open-plan office, causing heating and cooling in the same space. In a ddition – having uncovered a large chiller running all day in the middle of winter – the system found that a 2kW personal electric heater was fighting the centralised cooling plant, kicking in an enormous chiller and central pumping, which kept the cold water circulating.

Joe Short, chief executive officer at Demand Logic, says: ‘This discovery was possible because the system monitors each of the many hundreds of ceiling-mounted air conditioning outlets in the building, and it was able to identify the one demanding unusual amounts of cooled water.’ The system is different from a metering approach, however, where the objective is to get a full picture of energy flow. ‘That’s not our end game,’ says Short. ‘What we’re asking is whether energy is being wasted unnecessarily. It is often easier to find and fix an energy wastage than to measure it.’

[edit] Closing the performance gap

Funded by Innovate UK – formerly the Technology Strategy Board – the £900,000 project is a huge learning resource. Randall says the generation of performance data – which will be as open as possible – will provide industry, academia and government with a pool of data for further analysis. He adds: ‘We’re streaming about three million data points a day, from lots of buildings, so there’s huge potential to carry out data analytics with our partner, London South Bank University.’

By working with Verco and EDSL Tas, the continuous data stream would also be used to verify – and add accuracy to – building physics models to help close the performance gap. Demand Logic intends to develop methodologies for identifying the energy savings made that are compliant with the international performance measurement and verification protocol – a globally recognised approach to claiming savings achieved. The aim is to include approaches, allowed by the protocol, that use calibrated building physics models.

Alan Jones, managing director of EDSL Tas, says the system is a continuation of the BIM (building information modelling) process. A design model, which has gone through compliance calculation, is changed into an expected operational performance model. A second copy of that model will be taught by the system about what the building is doing, identifying how and why variations occur. The client would know that the compliance energy calculation is different from what they are actually seeing, whether that is because they are using it for twice as long, or running it poorly. ‘These questions could be answered, and we would be able to look at projecting future implications of these discrepancies,’ says Jones.

Moreover, the system would potentially provide an evidence base for building owners to invest in higher-cost energy efficiency measures. Randall says: ‘If a landlord decides to make a big change to the building system, they would have a calibrated thermal model with which they can get a better insight into the effect it would have.’ This data could then be used in the design of another building with similar characteristics.

Dave Worthington, managing director of Verco, says getting to the bottom of the discrepancy between design-stage assumptions and reality is challenging – it could be because of weather effects, occupant behaviour, or changes in the kit that was installed or its parameters. ‘It really is a “big data” challenge, which needs the collaboration of the different brains around the table that we have on this project,’ he says. ‘The performance gap is becoming a hotter and hotter topic. Significant progress has been made in the domestic sector, but the non-domestic sector is more difficult. ‘This gap exists because of a lack of access to credible data that people trust and understand, and can make business decisions on the back of them.’ Verco will look to feed the results back to government so it can ensure its energy-related targets are on track and it is coming up with evidence-based policies that incentivise the right behaviours in non-domestic buildings.

[edit] An exciting future

By applying Demand Logic – and having better access to data – Randall says there is potential for FM companies to take more ownership of their energy management programmes and performance risk . The bigger picture is to optimise operational performance of buildings and help close the performance gap. ‘If we succeed, we should make a profound difference to the tools that building services engineers have to hand to deliver buildings that actually work,’ says Randall. ‘It gives us a potentially interesting future, where we are looking after our buildings long-term, and getting value into our businesses.’

[edit] Automation

Demand Logic extracts trend information from the BMS to interrogate the issues. In-house specialists can sift through potential problems, but the team wants more and more of the abnormality detection to be done automatically. Short says: ‘Even a medium-size office has 20-30,000 elements that could be queried, and – over the years – several groups of engineers would have come and labelled them in a different way. ‘We are developing a simple machine learning technique to explore these text labels, because we’re faced with all of this data in an unknown building, and we’ve got to turn it into something that the owners and managers of that building know is their plant.’ Short says the most powerful thing Demand Logic provides is the simple element of being able to add comments on every query view. ‘It’s a place where you can air your hunches and suspicions about a building, in a free and uncriticised way.‘We’re getting beyond the technology, and learning about what context people are working in, professionally – what incentives are driving them, what culture they are working in, what politics are goingon, and what set-up encourages them to act.