THE FINAL TOLLGATE
DefineBOD Waste ReductionProblem Statement: Current operations produce BODwaste levels averaging 3864 mg/L - 1546% higher thanthe upper spec limit from Columbus City Utilities.
Goal Statement: At least 50% reduction in BOD by3/15/10 and the creation of a sustainable systemcapable of reaching a long term target of 95% reduction.
Expected Benefits: Dramatic reduction in solid waste entering floor drains Increased throughput at various plant processes Green initiative for the organization $195,000 in direct, hard cost reductions
DefineBOD, which stands for Biological Oxygen Demand, is ameasure of water quality and relates to the amount ofoxygen required by bacteria to breakdown solid wastes inwastewater. The City of Columbus requires that BOD notexceed 250 mg/L; otherwise extra strength surcharges areapplied with the possibility of more severe consequencesincluding full plant closure. Given this, Kahiki's support forthe environment and close to $204,000 in surcharges in2009, Kahiki chartered a six sigma project to dramaticallyreduce BOD in exiting waste water.
Creation of a good working relationship with thecustomer (City of Columbus) was paramount at this stagein the roadmap and served as a great starting point.Columbus City provided the team with valuableinformation about BOD and noted that "BOD starts as foodwaste entering floor drains". The team created processmaps and multiple SIPOC maps to help isolateopportunities for waste to hit the floor and understand theflow of food waste throughout the facility.
Food waste is created at various productionprocesses including raw chicken processing, fryingprocesses, packaging and dishwashing. Food waste that isnot thrown away or recycled eventually makes its way tothe plant floor, and is sprayed clean into one of 50+ plant
SIPOC for BOD Waste ReductionInput
- ~inentory :T,.........n:AIIt'~I ~7Ibwimla'ltory : ~~ . rtaIS ~
~-:- !p=,~~~ ,; I "'
THE FINAL TOLLGATE
MeasureBOD Waste ReductionProcess Capability: Process capabilitycharts were created for BOD and TSSmetrics. Both metrics clearly notcapable of meeting customer spec withcurrent waste removal system.
Measurement Plan: Success in Measurehinged on the execution of a detailed andefficientmeasurementplan.
Process capability of BOOQllwIilIlOM 6IJsed on \Velbull Dl5lJiI;lol.tIon Model
IS-~ II,~ 11US!. ;:5) II50=='-_ lUU1 II=-" ~ 11Sew ~ IIn...;,x ~ II1=::::1 ::
"" f:: ~" I" I
*Many of thesemetrics hadnever beenrecorded somuch time wasspent developingoperationalmethods of datacollection
Solidwaste Found at Drain Basket to captureSepa...- time series data Ibs
Willtells us how much solid waste enterssepa...-from sump
BOO. lSS a>ncentration more frequent samp!"_ floomCity
Cityof Gahanna.w_Js- readi"ll our own meterConsumption -water consumption
Gallons or Gal/hr.min,sec
IbsAmount of waste Sampledrain baskets
entering drains at various and wei&hsolids afterplant locations 2nd shift
SolidWaste found atSump
Solids), and TKN (Total Kjeldhal Nitrogen) would still bemonitored as the team hypothesized these metrics may becorrelated with BOD.
MeasureOne of the critical challenges in the Measure phase wasthat of data collection. Kahiki tracked extra strength levelsback to late 2007 however due to a major system change,data from 2008 and on could only be used asrepresentative of the current system.
The standard sampling schedule for Kahiki was 3site visits per year, each recording 4 samples and takingthe average for all metrics. This gave the team 24 datapoints over 2 years for BOD. For sake of future statisticalanalysis, the team pushed for monthly site visits togenerate more data and also allow for smaller turnaroundtimes for future pilot tests. This came at an extra cost,however in retrospect this decision point was critical to thesuccess of the project in the timeframe given.
Given the small amount of data, normality testswere performed and both BOD and TSSwere found to benon-normal. By fitting appropriate distributions andtransforming the data we were able to characterize bothmetrics in terms of their statistical capability. As expected,both metrics were highly incapable of meeting
customer spec limits.A demand analysis was performed to understand
what "pulls" existed on the waste removal system. As theteam began crystallizing potential process inputs, the needfor a detailed and accurate measurement plan becameapparent. The team turned potential inputs into questionswhich formed the underpinnings ofthe measurement plan.
The team gathered data that included historicaland current production volume, water and sewerconsumption, Ibs of waste entering various plant drainbaskets, solid wastes found at the sump and interceptorlocations as well as data dealing with interceptor/sumppump frequency. As the measurement plan beganexecution, more data began rolling in for our primarymetrics.
Another integral part of data collection was that ofsample power. In order to eventually make claims aboutthe data and potentially use hypothesis testing, the teamworked backwards from an appropriate sample power toarrive at the appropriate amount of data to collect.
One challenge that arose during data collectionwasthat of data collection visibility. Some of the morecritical data points needed came from 3rd shift which isdesignated solely to cleanup. Developing closerelationships and accountability was critical to ensure dataintegrity for these data points.
MaY/June 20:"0 OSU/ISE 3
THE FINAL TOLLGATE
BOD Waste Reduction
5 Why Analysis: This tool helped theteam get to the root of the problem.
Why is there a Gap between Current State andFuture State?
Why aren't BOD levels in customer spec?
DWhy are BOD levels too high?
DWhy is there too much waste in city sample?
DWhy does too much waste flow through Kahikrs
Waste Removal System?
Why does waste enter drains at various prod. points?
DWhy is waste created?
Throughput Yield not 100% at every process
Benchmarking: Almost every food processing plant deals with residual foodwaste. Rather than reinvent the wheel, the team did significant benchmarking.
Plant Floor Drain "Hot Spot" Chart: This depiction shows a heat map of solidwaste entering floor drains. This mapping allowed for strategic deploymentof solution elements targeting waste problem areas in the plant.
AnalyzeCharacterizing the look, feel and performance of apotential future state was critical during the beginningportions of Analyze, The team examined best in classindustries with regard to BOD and also examined Kahiki'sown performance to establish a potential future state forthe system. From this, the team did a gap analysis tounderstand what specifically needs to happen to certaininputs in order for the system to eventually achieve thisfuture state,
Once the gap analysis was performed, it becameapparent that certain inputs into the system were morecritical than others and also farther away from optimalthan others, The team did multiple cause and effectmatrices and FMEA's to examine potential inputs and theireffect on primary metrics as well as secondary upstreammetrics. A fish bone diagram that was performed in thelatter stages of measure helped get ideas rolling regardingpotential inputs into the system.
Using the data collected from the measurementsystem allowed for the construction of a drain "hot spot"heat map. This map uses a color scheme to characterizevolume of waste found in various plant floor drains. Thosedrains with red spots (9 drains) account for 70% of thetotal amount of solids found in all 50+ drains, This tool
70% of WasteCollected from:
25% a/WasteCollected from:
G.5% of WasteCollected from:
5 Worst Drains:37,9,20,21,2
~ 3 '~reQs~' toTarget: Raw
Process BaggerArea. Dishwash
allowed for an intensive look at potential problem areas inthe plant with regard to waste creation. It would alsoallow for strategic implementation of potential solutionswhich would target problem areas.
Significant benchmarking was done on multiplecompanies in the food processing industry. Because themajority of food processing facilities have to deal with theissues of waste creation, the team wanted to know aboutbest in class practices and also root cause identification.One salient example came from a company similar toKahiki in terms of volume and products made which servedas a tremendous resource during this phase of OMAIC.
Regression analysis was performed to examinestatistical relationships between inputs and a 5 whyanalysis was performed using the expertise of the team,subject matter experts and also benchmarked companiesto investigate root causes. This combined with the resultsfrom the C&E Matrices, FMEA and benchmarking allied tothe root of the problem: Waste creation at productionprocesses.
The regression results also helped the teamunderstand relationships involving water consumption,production volume and sump pumping practices. Theteam also did significant biological research to identifyspecific substances with regard to their invididualcontribution to BOD.
i '.?yJJwv' 2010 OSU/ISE 4
THE FINAL TOLLGATE
BOD Waste Reduction
Improved Solid/Liquid Separation: Inefficient solid/liquidseparation from FMEA led to a pilot test using a moreefficient separator.
Reducing Waste Creation: Various projects were initializedto increase throughput and decrease the amount of wastecreated at the source.
Employee Training: Simple but effective employee trainingwas implemented rega