Regulation 3.2 AUSTRALIA PATENTS ACT 1990 COMPLETE SPECIFICATION FOR A STANDARD PATENT ORIGINAL Name of Applicant: Schneider Electric (Australia) Pty Limited Actual Inventors: Adam Turner Colette Munro Michael MacKenzie Address for Service: C/- MADDERNS, GPO Box 2752, Adelaide, South Australia, Australia Invention title: SYSTEM AND METHOD FOR INDENTIFYING ENERGY OVERCONSUMPTION The following statement is a full description of this invention, including the best method of performing it known to us.
2 PRIORITY DOCUMENTS The present application claims priority from: 2009902376 entitled "Energy Management System" and filed on 25 May 2009. The content of this application is hereby incorporated by reference in its entirety. 5 INCORPORATION BY REFERENCE The following publication is referred to in the present application and the content is herby incorporated by reference in their entirety: US Patent 7409303 entitled "Identifying energy drivers in an energy management 10 system" to Power Measurement Ltd issued on 5 August 2008. FIELD OF THE INVENTION The present invention relates to energy management. In a particular form the present invention relates to a system for identifying the causes of energy overconsumption in a 15 process. BACKGROUND OF THE INVENTION In the current economic climate industry is facing increasing pressure to use energy more effectively. Companies need to do the same or more while using less. Additionally large 20 energy users also subject to regulatory pressure to report on energy usage and identify energy efficiency opportunities such as through schemes such as the Australian Energy Efficiency Opportunities. Participants must identify, evaluate, and reporting publicly on cost effective energy savings opportunities, with participation being mandatory for large energy users (which as of 2009 equates to approximately 210 companies). 25 Whilst industrial users, such as those in the mining, minerals and manufacturing sectors, are under pressure to reduce usage, they face several hurdles in reducing their energy usage. In order to reduce their energy usage, energy users must have a clear understanding of their current energy use. In the past such users have not been able to obtain the required 30 understanding due to existing energy management systems, if present at all, being insufficient for the task. Prior art systems typically suffer from deficiencies such as collecting insufficient data, or the data collected is not granular enough, or the collected data may be isolated and not made generally available. 35 Also many users lack the ability to either accurately forecast their usage, or if they can forecast their usage, then they are often unable to identify why they have used more or less 3 Also many users lack the ability to either accurately forecast their usage, or if they can forecast their usage, then they are often unable to identify why they have used more or less energy than usual. Large users are often charged different rates depending upon the amount of energy they use and are often required to forecast their energy usage to utility companies or 5 dedicated energy providers. Under forecasting or over forecasting energy use often results in financial penalties. Typically the challenge of reducing and reporting energy usage has been handled by electrical departments. These departments have largely responded by investing in meters and 10 monitoring software to report on energy usage. Energy monitors, controllers and associated software can be used to log how much energy an individual device is using. Such systems can also be used to show usage trends or to issue a warning when usage exceeds a predefined threshold (for example when a over usage penalty will be imposed). Energy departments have typically focussed on efficiency gains such as installing more efficient lighting or installing 15 lighting control systems (e.g. from 24 hour lighting to lighting only when required), switching to more energy efficient air conditioning units, and replacing fixed speed drives with variable speed drives. Whilst such approaches have resulted in energy savings, there is a limit to what can be achieved with efficiency gains. Energy efficiency gains and device monitoring indicates a piecemeal approach to the problem due to a lack of understanding of their energy 20 usage or the cause of energy overconsumption. There is thus a need to enable users, and in particular industrial energy users to identify causes of excessive energy consumption or at least to provide them with a useful alternative. SUMMARY OF THE INVENTION 25 According to a first aspect of the present invention, there is provided a method for identifying causes of energy over consumption in a process including: receiving one or more items of energy usage data for a process; receiving one or more items of contextual metadata associated with the process; receiving one or more threshold energy usage levels for the process; 30 identifying one or more energy overconsumption events by comparing the received energy usage data with the relevant threshold energy usage level; associating energy overconsumption data with each of the one or more energy overconsumption events identified, wherein the energy overconsumption data includes the amount of energy consumed in excess of the relevant threshold, and the energy 35 overconsumption time period; 4 analysing each energy overconsumption event to identify one or more possible causes for the energy overconsumption event, wherein the analysis uses at least one of the one or more items of contextual metadata associated with the process. 5 The energy usage data may be received from individual devices, power meters monitoring several devices, and/or SCADA systems monitoring the process and may be in the form of individual data points, aggregate use over a time period, a time series etc. Contextual metadata may be received from plant or business information servers. Threshold energy usage data may be received from a user (via a user interface) or from computational model, such as 10 one which forecast the expected usage based upon historical data and/or current process inputs. The energy overconsumption data may further include data on the time structure of the energy overconsumption, and analysing each energy overconsumption event includes identifying one 15 or more structural features in the energy overconsumption event and associating on or more items of the contextual metadata with each one of the one or more structural features. Alternatively the energy overconsumption data may further include data on the time structure of the energy overconsumption, and analysing each energy overconsumption event may 20 include: splitting the energy overconsumption event into two or more sub-events; and separately analysing each sub event to identify a cause of the sub-event. The step of splitting an energy overconsumption event may be based upon one or more 25 energy usage levels, or one or more times. Further each item of contextual metadata may have one or more times associated with the item, and the step of splitting an energy overconsumption event may be based upon a time associated with an item of contextual metadata. 30 The method may further include identifying one or more structural features in the energy overconsumption event, and splitting the energy overconsumption event is based upon one or more of the one or more structural features identified. Analysing each over energy consumption event may further include: 35 associating a time period with each item of contextual metadata; 5 selecting a subset of contextual metadata items based upon contextual metadata items having an associated time which overlaps with, or is contained in, the energy overconsumption time period for the energy overconsumption event; identifying a cause of the energy overconsumption event based on the subset of 5 contextual metadata associated with the energy overconsumption event. The one or more items of contextual metadata may be classified into one or more predetermined descriptive classes, and wherein analysing each energy overconsumption event may further includes associating one of the one or more predetermined classes with the 10 energy overconsumption event and the one or more predetermined descriptive classes may include people, process and materials. The one or more items of contextual metadata may also include metadata relating to process stage, equipment in use, equipment status, work crew, years of experience of work crew, shift, 15 grade of input material, amount of input, amount of output and/or environmental conditions. Analysing each energy overconsumption event includes performing a root cause analysis using the contextual metadata. The process may also includes one or more sub processes, and receiving one or more threshold energy usage levels includes receiving a threshold level for 20 each of the one or more sub process forming the process. Further each item of contextual information may be associated with one or more sub-processes. The one or more threshold energy usage levels may be dynamically forecast thresholds, and these may be forecast using a model based on predefined energy drivers for the process, and 25 the value of those energy drivers in the current process. The method may further the step of including calculating one or more performance indicators for the process, wherein the performance indicator is based upon the amount of over consumed energy from the one or more energy overconsumption events identified. 30 According to a second aspect of the present invention there is provided a computer usable medium having a computer readable program code embodied therein, said computer readable program code adapted to be executed by a computer to implement the method for identifying causes of energy over consumption in a process according to the first aspect. 35 6 According to a third aspect of the present invention there is provided a system for identifying causes of energy over consumption in a process, the system including a processor coupled to a memory, including: a plant information module for storing one or more items of contextual metadata 5 associated with a process; an energy usage input module for receiving one or more items of energy usage data relating to one or more devices used in the process; an energy usage threshold estimation module for determining one or more threshold energy usage levels for the process; 10 an energy overconsumption module coupled with the plant information module, input module, and the an energy usage threshold estimation module, wherein the energy overconsumption module is for identifying one or more energy overconsumption events by comparing the received energy usage data with the associated threshold energy usage level, and associates energy overconsumption data with each of the one or more energy 15 overconsumption events identified, wherein the energy overconsumption data includes the amount of energy consumed in excess of the relevant threshold, and the energy overconsumption time period; and an analysis module for analysing each energy overconsumption event to identify one or more one possible causes for the energy overconsumption event, wherein the analysis uses 20 at least one of the one or more items of contextual metadata associated with the process. BRIEF DESCRIPTION OF THE DRAWINGS An illustrative embodiment of the present invention will be discussed with reference to the 25 accompanying drawings wherein: FIGURE 1 is an energy overconsumption analysis system according to an embodiment of the present invention; FIGURE 2 is a flowchart of a method for identifying causes of energy overconsumption in a process implanted in the energy overconsumption analysis system of Figure 1 according to an 30 embodiment of the invention; FIGURE 3 is an example of a report of an energy overconsumption event according to an embodiment of the present invention; FIGURE 4 is another example of an energy overconsumption event according to an embodiment of the present invention; 35 FIGURE 5 is another example of an energy overconsumption event according to an embodiment of the present invention; 7 FIGURE 6 is flowchart of the identification of the energy overconsumption event of Figure 5; FIGURE 7 is an example of splitting the overconsumption event of Figure 3 into sub-events; and FIGURE 8 is a database design diagram to support recording of an energy consumption event 5 according to an embodiment of the present invention. DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS OF THE INVENTION Previous energy management systems have typically focussed on replacing existing systems with more energy efficient systems, or providing more detailed reporting of energy use by 10 equipment or location (e.g. area or substation) to obtain usage trends. Some systems allow forecasting of the amount of energy that a user should expect a process to consume based upon historical data or known energy drivers so that alerts can be raised when usage exceeds an expected level. However replacing inefficient equipment, performing accurate forecasting, and monitoring of usage trends has only partially solved the problem of identifying new 15 opportunities to reduce energy use. The present invention addresses this issue by providing managers not only with information on energy overconsumption events, and in particular information on the structure of the energy overconsumption events, but also contextual metadata on the process. Armed with this combination of information managers can perform analysis to identify possible causes of energy overconsumption events, and thereby perform 20 remedial actions to prevent or reduce the likelihood or significance of future overconsumption effects. To assist in understanding the invention, embodiments will now be described in the context of industrial process. Industrial processes may be batch processes, continuous processes or some 25 form of hybrid. Typically an industrial process may be divided into two or more sub processes or parts. These may be performed by different pieces of equipment, or by different groups of equipment. Work crews may work on specific pieces of equipment and thus only be involved with a specific sub process. An industrial process (or sub process) may span several shifts, and so different sub-processes may be performed by different shifts. In some cases 30 different sub processes may run concurrently, and may each contribute to overall energy usage. Sub processes may be performed in different locations. Some pieces of equipment, such as lighting or air conditioning may be common to the entire process, or several sub processes. Different sub processes may use different input materials, and output of one sub process may be an input to another sub process. Business/planning systems will typically 35 record much of this metadata associated with a process, along with metadata relating to the company and the industrial site. However in the past such the information contained in such 8 systems has not been regarded as relevant to the problems of identifying new energy saving opportunities, or identifying causes of excessive energy consumption. As such information contained in such systems have not typically been made available or used by energy management systems. 5 Thus one aspect in identifying new energy saving opportunities or causes of excessive energy consumption, that in the past has been missing, is the context of the energy usage. Without knowledge of process context information, or contextual metadata around energy consumption, it is difficult to understand what caused energy consumption to exceed a 10 threshold target or benchmark (assuming one has an accurate forecast in the first place). Without this information, the focus has generally been put on improving equipment energy efficiencies, which can yield good results, but there is a limit to such approaches. By focusing on the operational issues that cause excess energy usage and the process as a whole (rather than a single piece of equipment) energy savings opportunities can be identified. 15 Another aspect includes capturing energy overconsumption events. In the following description energy overconsumption events may also be referred to as excess energy usage events, or more briefly as energy events. Capturing the energy overconsumption event, that is information such as the time period during which energy usage exceeds a relevant threshold 20 energy usage level and the actual amount of energy over consumed (that is the amount in excess of the relevant threshold for that part of the process), provides information, which in combination with contextual metadata can be used to reveal detail on possible causes of the energy overconsumption event. That is by understanding in detail what the process is currently doing, a manager is better able to determine if they are over-consuming, and if so, 25 why and by how much. This approach switches the focus from how efficient a single piece of equipment is to how energy effective the process is. In the past mangers may have spotted an under or consumption event, but firstly lack the tools to quantify the amount of energy over consumption and secondly, without contextual information, no cause could be identified. 30 By way of example consider two cars that have the same energy efficiency rating (e.g. 1 5L per 100km). In the first case this may have been measured whilst the car was going up a steep hill, whilst the second case may have been measured whilst the car was going down hill. With this contextual information it is possible to identify that the first car is more energy efficient than the second car. In another example a process includes filling a tank for a mixing 35 operation. If an inexperienced work crew overfill the tank, then when mixing begins it will overflow, and pumps will need to be used to dispose of the overflow which consume 9 additional (or excess) energy. In this case the root cause is inexperience of the work crew which lead to the overflow event which required pumps to be turned on. Using more efficient pumps may reduce the amount of energy used in such a case, but providing better training to the work crew will prevent the overflow event which necessitated the use of the pumps. 5 However in order to identify this root cause, the managers need access to contextual metadata such as which work crews worked on the filling and mixing stages and their respective level of experience. The present invention relates to a method for identifying causes of energy over consumption 10 in a process using contextual metadata (information) associated with the process. Illustrative embodiments of the present invention will now be discussed. It is to be understood that the problems faced by these industries applies to industry in general, such as mining, mineral processing, manufacturing, logistics, warehousing, distribution etc. One or more items of energy usage data and threshold energy usage levels (i.e. expected or target amounts) may be 15 received by the system for each piece of equipment, or for each sub process, or for a substation or area of an industrial site. Contextual information may be received and each item of contextual information may be associated with one or more sub-processes. The system also allows quantification of the amount of energy overconsumption for an identified energy overconsumption event. That is the amount of energy used in excess of the relevant threshold 20 energy usage level associated with each identified energy overconsumption event (e.g. the threshold or thresholds applicable at the time of the energy overconsumption event). Combining such information with contextual information enables the causes of energy overconsumption events to be determined, as well as enabling the estimation of performance indicators. Continuous improvement approaches, such as root cause analysis can be 25 performed to identify underlying causes of overconsumption events using the contextual metadata associated with an energy overconsumption event, thereby leading to the ongoing identification of energy reduction opportunities. Referring now to Figure 1, there is shown an energy overconsumption analysis system 100 30 according to an embodiment of the present invention. A data processing server 120 may receive one or more items of energy usage information and threshold energy usage levels (also known as targets or forecasts) from energy readings 102, energy meters 104, Supervisory Control And Data Acquisition (SCADA) modules 106, a historian module 108 and/or an energy forecast module 110 via network 112 which may be a wired or wireless 35 network, and may include the internet, or access to third party networks or systems. Thus the energy usage data may be received from individual devices, power meters, and/or SCADA 10 systems monitoring the process and may be in the form of individual data points, aggregate use over a time period, a time series etc. The data may be be received from equipment actively involved in the process or from equipment in an idle state or not currently processing materials. The historian module 108 may contain archived information on past energy usage 5 by individual machines, areas, substations, sub processes or processes, etc. The data processing server includes connectors/integrators 121 to receive and process this information. This information may also be provided to an energy configuration module for recording the energy configuration of the system, or the connectors/integrators may receive information on the energy configuration which may be provided to other modules within the system, or 10 integrated with other data, as required. The connectors/integrators 121 process the received data to produce energy readings 123, that is, energy usage data for the process. The energy usage data (energy readings) may be from individual pieces of equipment, a sub section or section of an industrial site which may 15 include several pieces of equipment. The connectors/integrators receive the energy usage data associated with the appropriate source, such as by use of IP addresses or other identifiers including in the data. A lookup table or database table (or tables) could be utilised to map the device or source identifier to a record containing more detailed information on the source (e.g. name, location, calibration information, device characteristics, etc). The 20 connectors/integrators may be software modules or a combination of software and hardware. For example the connectors/integrators could include a plurality of low level software modules each of which is in control of a network interface card or communications device, and is responsible for sending and receiving data. A higher level supervising software module could communicate with the low level modules to send and receive information. Further the 25 higher level supervising module may perform processing and integration of received data for use by other parts of the system (e.g. summarisation, aggregation, curve-fitting, reformatting etc). The energy usage data may be in the form of a plurality of energy usage data points. The 30 energy usage data points may be individual meter readings (instantaneous usage) or aggregate (cumulative or integrated) usage since the last reading, or aggregate usage over a predefined period (e.g. 1 minute). The energy usage data points may be provided as a time series of measurements relating to real time, or near real time measurements, or aggregate or summary usages amounts. Data points may be periodic in nature, e.g. every second, minute, 15 minutes, 35 hourly or some other time interval, which may be predetermined so that the data points implicitly form a time series. Alternatively the data points may be irregular in time, wherein 11 each data point includes the measurement, time of measurement and optionally further details regarding the measurement or nature of the usage data (e.g. location, equipment, nature of measurement, etc). 5 The connectors/integrators 121 may also process the received data to produce approximations or estimates of energy usage data for the process. For example the energy approximation module 122 may use real time process information and device specifications to estimate energy usage using a stored formula. For example a conveyor may be known to be operating at a speed x and the weight on conveyor may be known to be w, therefore an estimate of 10 energy consumption, c, based upon the design or conveyor specifications may be calculated using a mapping function: c=f(x,w). The connectors/integrators 121 also process the received data to produce threshold energy usage levels or energy usage forecasts. This may be provided via the energy forecasts module 15 110, or from historical data provided by the historian 108. A threshold energy usage level may be a static level or a dynamic level (i.e. changing with time or sub process). The energy forecasts module 110 could be software implementing the system described in US Patent 7409303 entitled "Identifying energy drivers in an energy management system". US Patent 7409303 describes a system which can dynamically forecast the required energy usage of a 20 process using a computer model which uses the values of known energy drivers. Typical drivers are the amount produced (tonnes out), amount of input (tonnes in), quality of the feedstock, processing time (hours). This system can also use input drivers to determine key energy drivers and thus develop a computer model to calculate or forecast future usage. A scaling factor may be applied to a forecast to account for variability in measurements. For 25 example a forecast could be increased by 10%, 20%, or some other amount as deemed appropriate to generate a threshold energy usage levels. Alternatively, if a cost penalty is imposed if usage exceeds a target level, then threshold energy usage levels could be defined as some percentage (e.g. 75%, 95%, 100%) of this target level. Multiple usage level thresholds could apply at the same time. Identifying energy overconsumption events will 30 require application of the relevant threshold based upon the time, and/or the current usage level. The relevant threshold will be the threshold that applies to the device, devices, sub process and will depend upon the exact system configuration and what thresholds are in place, such as whether multiple concurrent usage level thresholds apply. For example if usage has already exceeded 75% of the target (first threshold level), then a further energy over 35 consumption event may be generated when the usage level exceeds 95% of the target (the second threshold level). Both the energy readings module 123, the energy approximation 12 module 122, historian 108, and energy usage forecasts module I 10 may provide usage measurements, estimates, usage thresholds (forecasts) for individual pieces of equipment, or they may be aggregated to summarise the total usage for groups of equipment, a sub site, or the entire site. 5 Energy over consumption event identification/detection module 125 compares the received energy usage data with the relevant threshold energy usage level (this may be on the basis of energy configuration information). The relevant threshold energy usage level may be the threshold relevant for the particular sub process, piece of equipment, or point in time. As 10 discussed above their may be multiple usage thresholds of the process, sub process, piece of equipment, or point in time. For example a first threshold may exist for energy consumption exceeding 75% of a target level, and a second threshold for energy consumption exceeding 95% of a target level. In this case the relevant threshold for identifying a first energy over consumption event is when the energy usage exceeds the first threshold corresponding to 75% 15 of a target level. A second energy overconsumption event may be identified or defined when the energy usage exceeds the second threshold corresponding to 95% of a target level. Thus for identifying the first energy overconsumption event the relevant threshold is the first threshold (75%) and for identifying the second energy overconsumption event the relevant threshold is the second threshold (95%). Note that the second energy overconsumption event 20 could be obtained from splitting the first overconsumption event by applying the second threshold, or it could be separate event which is a subset of the first energy overconsumption event. Separate energy overconsumption events could be produced for energy in the range 75%-95%, and for usage above 95%, or the energy usage above 95% could be a sub event of the energy overconsumption event for energy usage above 75%. Further energy usage data 25 and thresholds could be provided in absolute terms, e.g. kW, kWh, or in a relative term such as energy usage per mass of input, such as kWh/ton. When an energy overconsumption event is identified a data capture module 126 captures information such as the time period (e.g. start and end times), and the amount of energy 30 consumed in excess of the relevant threshold, and associates this information with an energy overconsumption event which is stored in data repository 130. The association could be performed using a relational database, in which a unique event identifier is created, and then the captured information is stored with the unique identifier. In particular capturing and storing the structure of overconsumption events allows identification of features such as time 35 points when the amount of over consumed energy changes (step change) or the rate of overconsumption changes (e.g. change in slope). Such structural features can then be 13 compared with metadata information in subsequent analysis to assist in building a more complete understanding of what the process is currently doing and so assist in identifying possible causes of energy overconsumption. 5 The data or information captured could be processed and stored in various formats (and then associated with the event, for example by storing an event id in a database record). For example the time period of overconsumption could be succinctly stored via recording or storing the start and end time of the overconsumption event (i.e. time when the usage rises above and below the relevant threshold (or thresholds). Alternatively a start time and time 10 over threshold could be associated, or the start and end times may be extractable from the captured data, such as when the data corresponds to (time, usage level) data points, or a time series in which the start or end time is known, and the data points are regularly spaced in time (with a known sampling interval). In some cases the data may be summarised and represented parametrically, for example a polynomial could be fitted to the energy usage data 15 and used as a compact representation of the structure of the energy overconsumption event. Additional metadata may be stored or associated with the energy overconsumption event (also referred to as an energy event) to allow later identification of the source of the energy overconsumption event. For example this may relate to the equipment or location which 20 contributed to the energy overconsumption event, such as which pieces of equipment were in use and what processes were underway. The energy over consumption event detection module may be operated in real time, or near real time (i.e. when the process is operating), or data processing to detect energy overconsumption events may be delayed and performed at a later time. 25 A plant/business information server 140 provides contextual metadata associated with the process to an application server 150, which also has access to the data repository 130 used for storing energy overconsumption events. The plant/business information server, also known as manufacturing execution systems, may store information on downtime, production, 30 metrics, inventory, quality, planning, cost, and maintenance relating to a process or sub process. The application server enables a user, via a client application 170, to analyse each energy overconsumption event identified by the energy overconsumption detection module 125 using contextual metadata provided by plant/business server 140. The application server 140 may process received contextual metadata to associate some, or all of the received 35 contextual metadata with a specific energy overconsumption event. The contextual metadata may also be provided from the equipment via the connector/integrators. The contextual 14 metadata may include information such as process or sub process stage, equipment in use, equipment status, work crew, years of experience of work crew, shift, grade of input material, amount of input material, amount of output and/or environmental conditions. Other contextual metadata may be collected and used. This contextual data may be associated with all, or a 5 part, of the process. For example work crew A may work on sub process 1, whilst work crew B work on sub process 2, in which case work crews A and B are each only associated with a part of the overall process. Work crew C may be involved with the process from start to finish. 10 The application server, or the user may identify one or more one or more causes for an energy overconsumption event using the contextual metadata associated with the energy overconsumption event. A cause identified by the application server may be reviewed and confirmed or changed by the user. The application server may include a web services module 152 which enables interaction with a user via a web interface. The application server can 15 provide results and classified overconsumption events to a reporting server 160, which can use this information to produce reports 180. Referring now to Figure 2, there is shown a flowchart 200 of a method for identifying causes of energy overconsumption in a process implemented in the energy overconsumption analysis 20 system of Figure 1 according to an embodiment of the invention. The method may be implemented on a single processor, or using a distributed computing system. At step 210 one or more items of energy usage data for the process is received. The energy usage data may include a plurality of energy usage data points. The items may be received 25 from different equipment, or for different stages of the process . Further they may be received from equipment in an idle state or not currently processing materials. At step 220 contextual metadata associated with the process is received. Each item of contextual metadata may be associated with all, or a part, of the process. At step 230 one or more threshold energy usage levels for the process is received. Steps 210, 220 and 230 may be performed in any order, and 30 may be performed by the same processor, or by a variety of distributed processors in communication with each other. The received data may be provided by a variety of distributed systems. A single threshold for an entire process may be received, or a series of time adjacent 35 thresholds may be provided for the entire process. The threshold may be provided as a time series of threshold points. A single time period could also have multiple usage level 15 thresholds, for example a first (lower) usage threshold of 4MW and second (higher) usage threshold of 6MW. Whilst most of the discussion focuses on usage over threshold, it is to be understood that the system could be used to detect under consumption events, or for usage within or outside of a range defined by setting lower and upper thresholds. 5 The received data is then used to identify one or more energy overconsumption events 240 by comparing the received energy usage data with the relevant threshold energy usage level. Energy overconsumption data is then associated with an energy overconsumption event 250. The energy overconsumption data includes the amount of energy consumed in excess of the 10 relevant threshold, and the energy overconsumption time period (e.g. start time, and end time). Each energy overconsumption event is then analysed 260 using the contextual metadata associated with the energy overconsumption event to identify one or more one or more causes for an energy overconsumption event. This analysis uses at least one of the received items of contextual metadata associated with the process. The analysis could be performed in real 15 time, or some later time, such as after completion of the process. Analysing each energy overconsumption event may be performed in an automated, semi automated or manual (user driven) fashion. When identifying the cause of an energy over consumption event, data on other (past or present) energy over consumption events may be 20 combined or used. Automated analysis may be performed by a computer module which may implement various statistical, machine learning or data mining approaches to identify a cause. Statistical analysis could involve linear regression analysis and multivariate regression analysis, and the various relationships may then be inspected to assess the quality of each relationship fit (e.g. using correlation or regression coefficients). Machine learning 25 approaches could be supervised or unsupervised, and use approaches such as fuzzy logic, genetic algorithms, neural networks, expert systems etc. Semi-automated analysis may include a first estimate of the cause, or a list of possible causes produced by a computer module which is then subject to a review by a user. Alternatively manual analysis or investigation of the data may be performed by the user. The common thread is the use of 30 contextual metadata in addition to energy usage data. In addition to evaluating over consumption events, contextual metadata may also be used to review overall energy usage. Analysis of each energy overconsumption event may include performing a root cause analysis using the contextual metadata. Root cause analysis is a tool often used in continuous 35 improvement processes which attempts to identify the root causes of problems or events, rather than just identifying the immediately obvious effects. Root cause analysis includes 16 defining the problem, collecting data, identifying possible causal factors, identifying the root cause, and recommending and implementing solutions. Root cause analysis advocates collecting a wide range of data from different sources and examining as many possible causal factors as possible. This includes investigating what sequence of events leads to the problem, 5 what conditions allowed the problem to occur and what other problems surround the occurrence of the central problem (i.e. the energy overconsumption event). Identifying possible causal factors may include generating flowcharts, graphs, and cause and effect diagrams. This process may be iteratively applied to causal factors identified to assist in uncovering the root cause. This may include determining why a particular causal factor exists 10 and what is the real reason for the problem occurring. Once root cause analysis has been completed, the cause location, a cause code, classification, effect and any further comments can be stored. In some cases the root cause analysis may be performed in real time and may require subsequent validation by a manager. Corrective and Preventative actions can the be developed and implemented to address the root cause and thereby prevent, or at lease reduce 15 the effect of a future energy overconsumption event due to the identified root cause. For example in the tank overflow example described above, the root cause was an inexperienced crew, whereas the observed effect was an overflow event which required pumps to be turned on. In this context, root cause analysis typically may involve looking for 20 correlations between events or effects, and contextual metadata. For example one may look at years of experience of the crew working on a piece of equipment that used excessive energy or if backup machinery was being used due to machine failure. Also rather than focussing on a single event, events may be grouped together (e.g. is the same work crew, piece of equipment, or sub process responsible for most of the energy over consumption events) or 25 compared with previous events (such as when a more experienced work crew used a piece of equipment or performed a sub process). In another example, a process may require 5 machines to paint a car and there may be2 backup machines in case some of the primary machines are broken or otherwise unusable. 30 Every once in a while, there is a quality issue in the car, (e.g. missing door), in which the car is taken off the production line, a door is added, and instead of putting the car back onto the production line, the car is painted using the 2 backup machines. If just the 5 primary machines are working, or just the 2 backup machines are running, then there is no energy overconsumption. However if all 5 primary machines are running and the 2 backup machines 35 are running then even though each individual machine is not over-consuming energy, the process as a whole is. An energy overconsumption event could be prevented by restricting use 17 of the backup machines when all 5 primary machines are operating, or scheduling use when not all of the5 primary machines are operating. Having the contextual information of which machines are running allows the cause of the overconsumption to be identified and the focus switches from how efficient a single piece of equipment is to how energy effective the 5 process is. In another example the quality of the feed material for a mill in an ore processing facility is harder than typical. This leads to an increase in energy usage for the milling sub process as the mill has to work harder to grind the feed material to produce the same quality output as is 10 typically produced. In this case contextual information such as the feed material being from a different source, or that the results of laboratory testing indicates that the feed material was harder than normal may be available in a business/plant information system. Analysis of the over consumption event may identify that there was a difference in the input material compared to what was typically used, and so the quality of the feed material may be 15 identifiable as the root cause of the excessive energy consumption. The analysis process for an energy overconsumption event may include assigning a time period to each item of contextual metadata. This may be the time point at which a piece of equipment is turned on, turned off, or it may be a time period when a piece of equipment was 20 operating, or possibly a time period when the piece of equipment was operating in a particular mode (e.g. idle or full power). Other examples include the time period when a sub process was being performed (or start and end times), or the how long a work crew worked on the process. Next the analysis can select a subset of contextual metadata having a time which overlaps with, or is contained in, the time period defined by the start and end time of the 25 energy overconsumption event. This typically reduces amount of metadata that needs to be considered so that the next step of identifying a cause of the energy overconsumption event can be performed based on the subset of contextual metadata associated with the energy overconsumption event. 30 The contextual metadata may be classified into one or more predetermined descriptive classes such as people, process, or material. Analysis of each (or all) energy overconsumption event may include reviewing the energy usage as a function of class for the time period associated with the energy overconsumption event (i.e. start time to end time). For example if a particular sub process generated an energy overconsumption event, then the years of 35 experience of the work crew could be examined for the relevant time period (over consumption event) as compared to past operations where the sub process has been 18 performed. When a cause for the overconsumption event is identified one of the one or more predetermined classes may be associated with the energy overconsumption event. The above method enables identification of the causes of an energy overconsumption event. 5 The integration of a companies business/plant information system, or other systems, provide the necessary contextual information (e.g. crew, shift, feed material) to assist in determining the cause of energy overconsumption events. Referring now to Figure 3, there is shown an example of a report 300 of an energy 10 overconsumption event according to an embodiment of the present invention. The y axis represents energy usage, such as in units of kW or MW and the x axis represents time g.. The report illustrates a fixed threshold energy usage level 320 and the received (actual) energy usage data for a process over time 310. An energy overconsumption event 330 is identified. A summary report 340 is provided below the plot, and include the start time 341, the end time 15 342, the duration of the energy overconsumption event 343, the amount of energy consumed energy in excess of the relevant threshold (in this case static threshold 320) for the duration of the event 344 (shown in MWh), the assigned classification 345, the assigned cause 346, the shift 347 and the work crew 348. 20 Threshold levels may be predetermined, or calculated dynamically, and they may be predetermined, calculated, specified or forecast separately for each sub process forming the process. A predetermined threshold (whether for an entire process or a sub process) may be input or otherwise received from a user, or read from a configuration file. Additionally or alternatively, a threshold may be calculated based on received information, information in a 25 configuration file (or database) and/or historical data. This could further take into account process specific values, such as the source or quality of inputs (e.g. estimate of ore hardness) or outputs (required fineness) etc. For example a configuration file could contain a formula which takes into account quality measurements of the input and assign low thresholds for high quality inputs, and higher thresholds for poorer quality inputs (which may require more 30 processing, or at least more variable processing). In one embodiment dynamic thresholds could be calculated during the process. This may be based on information from earlier in the process, possibly in combination with input from a user or a configuration file. A dynamic threshold may be calculated/predicted/forecasted using a model based on predefined energy drivers for the process, and the value of those energy drivers in the current process, such as 35 that described in US Patent 7409303.
19 Further thresholds may be set based on some percentage of an expected amount. In this way users can investigate events leading up to an energy overconsumption event, or alternatively in cases where an energy overconsumption event was just avoided. 5 Referring now to Figure 4, there is shown another example of an energy overconsumption event 400 according to an embodiment of the present invention. In this case a dynamically forecast threshold 410 is used, and when the energy usage 420 exceeds the dynamic threshold, an energy over consumption event is identified 430. The vertical (y) axis represents energy usage (e.g. kW or MW) and time is plotted on the horizontal (x) axis. 10 Referring now to Figure 5, there is shown another example of an energy overconsumption event 500 according to an embodiment of the present invention. In this case a dynamic threshold 510 is again used to compare against the actual energy usage 520. A flowchart 600 of the identification of the energy overconsumption event of Figure 5 is shown in Figure 6, 15 such as that implemented in the system shown in Figure 1. The vertical (y) axis represents energy usage (e.g. kW or MW) and time is plotted on the horizontal (x) axis. As can be seen in Figure 5, a change in the process occurs at 502 and 508, and an energy overconsumption event occurs due to the sub process being performed 530 (highlighted in 20 grey). Figure 5 also illustrates the use of a start delay and stop delay which is further described in relation to Figure 6 to provide a more robust approach to identifying the start or end of an energy over consumption event. The event detector module 600 starts 602 in an off state 604. An event trigger 606, such as reception of new energy usage data, or expiration of a periodic timer (e.g. every minute), leads to a comparison of the actual usage with the 25 threshold or target usage 608. The event detector module 600 returns to the off state 610 if the usage was less than the threshold. If the usage was above the threshold then a check is made if a start delay is configured 612. If a start delay is configured the system enters a pending on state 614. When the start delay expires 616 a further test of usage over threshold is performed at 620. If the result is no, the system stays in the off state 604, as the usage was only 30 transiently above the threshold. If however no start delay was configured, or if the usage stayed above the threshold after expiration of the start delay, then system enters the on state 618. An event trigger 622, such as further data, or the expiration of the periodic timer forces a further check of usage above threshold 624. If usage is still above threshold the on state is maintained 618. If usage has dropped below the threshold a check is made if a stop delay is 35 configured 626. If no stop delay is configured then the event detector module returns to the off state 604. If a stop delay is configured, then a pending off state 630 is entered. After 20 expiration of a stop delay 632, a further check of usage above threshold is performed 634. If usage has returned to being above the threshold 636, then the module stays in the on state 618 (transient dip below). If the usage has stayed below the threshold, then the module returns to the off state 638. 5 The above procedure illustrates a robust approach to identifying energy overconsumption events, as transient deviations above or below the relevant threshold is ignored. An energy overconsumption record (energy record) is opened when the event first enters the on state. If a start delay is present the energy over consumption is calculated and includes the values 10 recorded during the start delay. Whilst in the on state or pending off state, the energy over consumption is periodically calculated, and the energy record is updated. Also when returning to the off state from the on state or pending off state the energy record is closed and the final energy overconsumption calculated includes the values recorded during the stop delay. Other robust variants are possible, for example usage data could be smoothed, or averaged using a 15 running or trimmed mean. The energy overconsumption record can then store or otherwise associate the start time, the stop time and the amount of energy above the threshold. This may be provided as numeric values, or a time series in which values represent the amount over consumed since the last time point. 20 In order to understand whether energy usage is effective, managers need to more fully understand the nature of the energy usage. In some cases this requires analysis of the structure of the over consumption event to identify features or time points when usage level changes. Such changes can be step changes or changes in the slope or rate of energy usage. These time points can be used to define sub events which can be analysed in more detail to identify root 25 causes of the usage level change. Some contextual metadata may have an associated time, and these times can be compared with these identified time points when usage level changed to select a set of contextual metadata that may provide metadata on the root cause of the overconsumption event. 30 Analysis of the structure of energy overconsumption event may include splitting an energy overconsumption event into two or more sub-events and separately analysing each sub event. This splitting may be performed automatically, such as by the event detection software module or a dedicated software module which looks for sudden changes in usage, peaks (local maxima or minima), or changes in slope. Splitting could also be performed semi 35 automatically, such as a human reviewed computer estimate of the splitting point. For example a representation of the energy overconsumption event could be displayed to a user 21 with possible split points marked and requiring confirmation by the user. In another embodiment, a user could specify splitting criteria which is then automatically applied to any overconsumption events. Alternatively the splitting process could be manual or user driven such as via a client application which provides a graphical display of the over consumption 5 event in which a user may select splitting points. For example a user could be shown a representation of the overconsumption event and they could click on a time point on the representation. Alternatively or additionally a time slider could be displayed and the user could slide this to the desired time point. Alternatively or additionally a user could be provided with a text box in which to enter a splitting time. 10 More generally event splitting may be performed on the basis of one or more predefined splitting criteria. This may include energy usage level criteria, producing horizontal splitting of the event in which an event is split into different usage levels, rather than just on a time basis(e.g. usage under 4MW, between 4 - 6MW and over 6MW). Additionally or alternatively 15 an energy overconsumption event may be split on the basis of time associated with an item of contextual metadata, also known as vertical splitting or time splitting, wherein sub-events are adjacent in time, and the amount of energy consumed in excess of the relevant threshold is split between the sub events. Alternative criteria may be based upon at least one item of contextual metadata such as when a particular sub process is being performed, or a particular 20 work crew is working, or a piece of equipment is in use. Referring now to Figure 7, there is shown an example 700 of splitting the overconsumption event 730 of Figure 3 into three sub-events. The threshold 710 is shown as is energy usage curve 720. The energy overconsumption event 730 begins at start time 732 and ends at end 25 time 738, and includes two split time points 734 and 736, The event is time split into sub event I from 732 to 734, sub event 2 from 734 to 736, and sub event 3 from 736 to 738. Separate analysis of the three sub events could then be performed. For example in sub event I, a tank may be overfilled by operating pumps at too high a flow rate leading to excessive use of energy and the first overconsumption sub event. Next, a mixing operation may begin, 30 but due to the excess liquid in the tank, extra energy is required to initiate mixing of the liquid in the tank, leading to the second overconsumption sub event. Third, the extra energy used in mixing leads to an overflow, and additional pumps are required to clean up the spilled liquid, leading to the third overconsumption sub event. 35 In some cases different causes for several sub events may be identified, and together these may indicate the root cause of the overconsumption event, which may actually be prior to the 22 overconsumption event. In the above case the root cause was the initial setting of a pump at too high a flow rate. With knowledge of the amount of energy overconsumption, key performance indicators (KPI) 5 can be calculated using the value of the amount of over consumed energy from one or more energy overconsumption events. Example KPI's include percentage Over Consumption Compared to Total Consumption and kWh/tonne, or excess kWh/tonne. Such indicators enable the effectiveness of energy usage over time to be evaluated. Such KPI's can be included in reports 180 to managers 10 The method may be implemented computationally using suitable computing languages, hardware and software platforms. For example the system could be implemented using C#, .NET framework and Microsoft SQL Servers. Various user interfaces including client server and web based interfaces could be provided. The system could be implemented as a 15 distributed system in which the various parts are communicatively coupled using wired or wireless means. Databases may be used to store various data such as energy overconsumption events (records), and causes, classifications, effects, and user comments. The system and method may be embodied in a computer program product or a computer usable medium (e.g. DVD, Flash disk, etc) which includes computer readable program code which is adapted to be 20 executed by a computer to implement any of the embodiments of the invention described above. In one embodiment an Energy Reporting Point item could be defined which is used to define or record capture conditions (e.g. thresholds to be used, etc) and splitting conditions along 25 with any start and stop delays to be used in identifying energy overconsumption events. An energy configuration item could be used to store default configuration details, such as standard fields to be used for an energy overconsumption event. An energy overconsumption event can also be defined with a set standard fields used to 30 record information on an energy overconsumption event. An example set of fields is shown in Table 1. The standard fields include various auditing fields (CreatedBy, CreatedDateTime, etc), energy overconsumption data fields (StartDateTime, EndDateTime, Duration, OverConsumption, Equipmentld, EquipmentType) and causal fields (Cause, CauseLocation, CauseLocationEquipmentld). The energy overconsumption event could be implemented as a 35 data structure or as a database record, or set of related records. Some fields may not be stored in the database record or data structure, and instead may be calculated on the fly from stored 23 values using stored procedures. For example users may be allowed to search in specific time periods and in which case stored values may be may clipped to the queried time period on the fly.. 5 Table 1: Standard Fields in energy overconsumption event. Id IsClipped CauseLocationEquipmentld HasAudit StartDateTime CauseLocationEquipmentType CreatedBy StartDateTime.Clipped Effect IsManaul EndDateTIme CauseLocation CreatedDateTime EndDateTime.Clipped Cause ConfirmedBy Duration Classification ConfirmedDateTime Duration.Clipped Comments IsConfirmed ObjectId ActiveWithin IsDeleted LastModified OverConsumption IsSplit Equipmentld OverConsumption.Clipped ClippedPercentage EquipmentType Figure 8 is an example of a database design diagram for an energy overconsumption event according to an embodiment of the invention. The design utilises 8 related database tables labelled EnergyDataSet 810, EnergyDataField 820, EnergyDataSetSplit 830, 10 EnergyDataComment 840, EnergyAudit 850, AuditLog 860, Equipment 870 and Field 880. The EnergyDataSet table contains generic information on the event such as those fields shown in Table I (excluding values which are calculated on the fly such as the clipped values). To enable recording of sub events, or split events, the EnergyDataField table is used to store various standard fields such as LastModified, EquipmentlD, EquipmentType, 15 CauseLocationEquipmentld, CauseLocationEquipmentType, Effect, CauseLocation, Cause, Classification and Comments. Each entry has a Fieldid which is obtained from the Field table which defines field ids and associated information. The EnergyDataSetsplit table is used for recording event splitting information and includes a setId key which is also stored the EnergyDataField to enable storing of information on subevents and their specific causes. The 20 EnergyDataComment table is used for recording comments, and the Energy Audit Table and AuditLog tables are for storing editing information. Finally the Equipment table contains information on pieces of equipment such as location, equipment name, etc. User interfaces could be provided to allow a user to graphically view, edit and store 25 information. Contextual metadata may be stored in plant/business information servers or manufacturing execution systems which store operational data relating to the process. A query engine can extract this data and a combined display of energy usage data and contextual 24 metadata can be provided to the user. The user can then zoom in or out, produce further charts and perform further queries in order to identify a possible cause, cause location, equipment ID and equipment type, which can then be stored with in an energy overconsumption record having the above framework. For example energy consumption over a specified time period 5 could be plotted as a function of work crew, equipment, or feed material grade. By integrating contextual metadata contained in plant/business information servers 140 with energy usage data, the system described herein allows one to optimise energy usage for an industrial plant. Such a plant energy optimisation system allows plant and production 10 managers to improve their production efficiency, performance and profitability. By allowing data to be aggregated and validated from many disparate sources, and by applying a continuous improvement methodology to analyse, understand and quantify operational inefficiencies, significant improvements in plant performance can be achieved through enhanced management of assets and materials. For example data can be captured in real time 15 and an operator can monitor energy usage and perform root cause analysis on any over consumption event to identify a cause and classify the event. Maintenance or operation supervisors could review all the over consumption events, and their causes and classifications on a daily basis. Corrective or preventative actions can then be initiated and reviewed on a weekly basis to determine the effectiveness of the corrective or preventative actions taken. 20 Monthly meetings can then be used to review overconsumption events, their causes and solutions, and further opportunities to reduce energy usage evaluated. This system enables managers to better understand their energy usage and to identify energy saving opportunities. Further the system facilitates a continuous improvement based approach 25 to achieving energy reductions and more efficient use of energy, leading to ongoing benefits. Consider the example of a mining and mineral processing site that requires approximately 1.8 MW of power. A dedicated power plant (run by a 3rd party) produces 2.7MW and therefore has a spinning excess of 0.9 MW. The company only pays for the 1.8MW but has to pay for maintenance and fuel for the entire 2.7MW. 30 If sufficient information was known about why the peak loads occur (for example the starting sequence for mills used to grind minerals), the spinning excess could be reduced, therefore reducing the fuel and maintenance costs. Further previous energy overconsumption events may also be the reason for the size of the spinning excess. A better understanding of their root 35 cause and prevention may also support the reduction of the costs associated with spinning excess. For example the power plant could be optimised to operate at 2.2MW, reducing the 25 spinning excess from 0.9MW to 0.4MW if the root cause is estimated to be the grade of ore. Alternatively the required spinning excess could be estimated based on analysis of ore samples prior to grinding operations. 5 The above system utilising combined capture and analysis of both contextual metadata, and energy usage data enables energy users to increase their understanding of their actual energy usage and enables energy users to identify causes of excessive energy consumption. Energy users can then put in place actions to reduce their overall energy usage, and thus generate considerable savings. Further some over consumption events may be caused by failing 10 equipment and thus this information could be used as a trigger to optimize maintenance programs. The system also allows real-time metrics to be produced which can also help to drive behaviour. For example kWh/ounce can be calculated and displayed to operators to given them information on how the plant is performing in terms of production and energy. Further by collecting accurate records on the causes of energy consumption above a target, 15 this data can be used to support capital expenditure required to change process equipment. Those of skill in the art would understand that information and signals may be represented using any of a variety of technologies and techniques. For example, data, instructions, commands, information, signals, and may be referenced throughout the above description 20 may be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof. Those of skill in the art would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments 25 disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed 30 on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention. The steps of a method or algorithm described in connection with the embodiments disclosed 35 herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. For a hardware implementation, processing may be 26 implemented within one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, micro controllers, microprocessors, other electronic units designed to perform the functions 5 described herein, or a combination thereof. Software modules, also known as computer programs, computer codes, or instructions, may contain a number a number of source code or object code segments or instructions, and may reside in any computer readable medium such as a RAM memory, flash memory, ROM memory, EPROM memory, registers, hard disk, a removable disk, a CD-ROM, a DVD-ROM or any other form of computer readable medium. 10 In the alternative, the computer readable medium may be integral to the processor. The processor and the computer readable medium may reside in an ASIC or related device. The software codes may be stored in a memory unit and executed by a processor. The memory unit may be implemented within the processor or external to the processor, in which case it can be communicatively coupled to the processor via various means as is known in the art. 15 Throughout the specification and the claims that follow, unless the context requires otherwise, the words "comprise" and "include" and variations such as "comprising" and "including" will be understood to imply the inclusion of a stated integer or group of integers, but not the exclusion of any other integer or group of integers. 20 The reference to any prior art in this specification is not, and should not be taken as, an acknowledgement of any form of suggestion that such prior art forms part of the common general knowledge. 25 It will be appreciated by those skilled in the art that the invention is not restricted in its use to the particular application described. Neither is the present invention restricted in its preferred embodiment with regard to the particular elements and/or features described or depicted herein. It will be appreciated that the invention is not limited to the embodiment or embodiments disclosed, but is capable of numerous rearrangements, modifications and 30 substitutions without departing from the scope of the invention as set forth and defined by the following claims.