Applications of the CUSUM Method to Energy Management
Speaker: Matthieu Frappe
[email protected]
Presentation content 1. About KYOTOenergy 2. CUSUM in a nutshell 3. Dynamic Process Control 4. Target Setting 5. Input Normalization 6. CUSUM Sampling Period 7. Applications to Energy Management
8. Conclusions
KYOTOenergy: Main activities • ENERGY ADVISORY SERVICES • Energy Audit • EMS implementation • Energy efficiency project design/ implementation
• CLIMATE CHANGE ADVISORY SERVICES • Emissions reduction project • Carbon GHG reporting (foot-printing) & compliance with regulations • Commercialization of carbon asset • Climate change policy study
• MONITORING & REPORTING SERVICES • CDM/VCS projects • Performance monitoring for acceptance tests, due diligence, third party verification
Slide 3
KYOTOenergy: Professional skills • Experts in a wide range of technologies ENERGY EFFICIENCY
• Energy audit of buildings and industrial clients RENEWABLE ENERGIES
• • • •
Biomass cogeneration Hydropower Wind power Geothermal
WASTE MANAGEMENT
• Composting of agricultural waste • Methane capture and reuse (biogas to power)
Slide 4
KYOTOenergy: Project references Geographical distribution Argentina, 1 Guatemala, 6 Uganda, 1 Tanzania, 1 Malaysia, 7 Indonesia, 4 Singapore, 7 Vietnam, 60
Thailand, 2 Philippines, 1 Laos, 1
Slide 5
Selective client list
Applications of the CUSUM Method to Energy Management
Slide 6
CUSUM in a nutshell •
CuSum : Cumulative Sum of data deviations from the average trend line.
Slide 7
CUSUM in a nutshell CuSum : Cumulative Sum of data deviations from the average trend line. Energy consumption
•
Capacity
Slide 8
CUSUM in a nutshell •
Then the deviations are cumulated over time, showing aggregated trends that are not obvious from raw data.
Slide 9
Dynamic Process Control •
Original purpose of the method, back in the 50s’
•
Generates exception reports from deviations on the spot
•
Comparable to Shewhart control chart:
Slide 10
Dynamic Process Control •
But instead of Shewhart control chart, CUSUM: •
Detects persisting deviations on the spot, due to its
integrating nature. •
Considers deviations from a multivariate reference trend line, enabling to cancel out the impacts of input
parameter fluctuations (capacity, weather, etc.), keeping focus on the efficiency of the system itself.
Slide 11
CUSUM for Target Setting •
A target should be: Specific, Measurable, Achievable, Relevant, Time-bound.
•
Realistic, so that the system members are really accountable for results and not discouraged.
•
Ambitious, so that opportunities are fully seized.
•
Consistently achievable over time, taking into consideration long term behaviors of the system.
Slide 12
Energy consumption
CUSUM for Target Setting
Capacity
Need to consider the time consistency of these points!
Slide 13
CUSUM for Target Setting • Draw CUSUM time-profile • Select the best sample period:
Slide 14
CUSUM for Target Setting
Energy consumption
• CUSUM results: Best achievable target
Capacity
Slide 15
Input Normalization •
Accountability of system manager for results: •
Many decisive parameters for the energy system
(capacity, weather, raw material quality, product quality, equipment, staff, etc.) •
Need to focus on what the energy system manager
can actually control. •
=> Need to cancel out known influencing factors.
Slide 16
Input Normalization: Capacity Capacity normalization Energy consumption
•
Capacity
•
Use of linear trend line ( Energy = a*Capacity + b)
Slide 17
Input Normalization: Weather •
Weather normalization •
•
Warmer air: •
Helps combustions and heat production
•
Reduces heat losses
Colder air:
•
Helps compressed air production
Slide 18
Input Normalization: Weather •
Example with heating load:
•
Calculate temperature difference between the outdoor air and final hot air to produce. (here T hot = 50ºC) :
•
Correction: ΔT% = (T hot – T air) / (T hot – Mean[T air]) Date Aug 2012 Sep 2012 Oct 2012 Nov 2012 Dec 2012 Jan 2013 Feb 2013 Mar 2013 Apr 2013
T air
ΔT= T hot – T air
ΔT%
29.6 °C
20.4 °C
76%
27.3 °C
22.7 °C
84%
26.6 °C
23.4 °C
87%
20.8 °C
29.2 °C
109%
15.1 °C
34.9 °C
130%
15.3 °C
34.7 °C
129%
19.8 °C
30.2 °C
112%
23.8 °C
26.2 °C
98%
23.3 °C
26.7 °C
99%
Slide 19
Input Normalization: Weather •
Normalization results:
Slide 20
CUSUM Sampling Period •
Find the minimum duration for a representative sustained operation: 2 weeks? 5 weeks? 9 weeks?
•
Discussion similar to “MCR” definition.
Slide 21
Application in Energy Management
Target setting
•Act
•Plan
4
1
CUSUM series
3
2
CUSUM target of previous year
•Check Dynamic control
• Do
Application in Energy Management • Holistic approach: • No need to know, model, and assess each sub-system component in a bottom-up approach. • CUSUM scans the overall system behavior’s history. • Time resolution of data:
• No need for the highest frequency. • Only need to match the frequency of the normalization inputs. (capacity fluctuations, weather change)
Slide 23
CUSUM: Summary • Principle applicable to any system, particularly energy consuming facilities. • Focus on overall system efficiency, extracting trends. • No need for detailed process understanding or modeling. • Possibility to filter out external factors, keeping the focus on the efficiency of the system itself. • What CUSUM can do: • Set targets at best sustainable levels (tough but achievable)
• Identify efficiency changes on the spot, and trigger early response.
adfdaadf
adf
Thank you Michel Buron – CEO
[email protected] +65 6248 4728 Arijit Paul – Regional Manager (Mekong Region)
[email protected] +84 4 3564 1940 Matthieu Frappe – Project Manager
[email protected]
