Abstract #228
Section: Extension Education
Session: Extension Education
Format: Oral
Day/Time: Monday 4:00 PM–4:15 PM
Location: 327
Session: Extension Education
Format: Oral
Day/Time: Monday 4:00 PM–4:15 PM
Location: 327
# 228
Maximizing income over feed cost by grouping cows with mixed-integer programing.
Y. Wu*1, V. Cabrera1, R. Shaver1, 1University of Wisconsin-Madison, Madison, WI.
Key Words: nutritional grouping strategy
Maximizing income over feed cost by grouping cows with mixed-integer programing.
Y. Wu*1, V. Cabrera1, R. Shaver1, 1University of Wisconsin-Madison, Madison, WI.
Study objective was to maximize income over feed cost (IOFC) by mixed-integer programing (MIP), an optimization engine with objective function: max(IOFC) = Σ(g1i × MYi × P) − Σ(g1i × DMIi × Path(g1i × NECi) × C1) − Σ(g1i × DMIi × Pbth(g1i × CPCi) × C2) + Σ(g2i × MYi × P) − Σ(g2i × DMIi × Path(g2i × NECi) × C1) − Σ(g2i × DMIi × Pbth(g2i × CPCi) × C2), s.t. gji∈(0,1), g1i + g2i = 1, where j is group, i is cow. Milk yield of cow i (MYi) is constant. NECi, CPCi and DMIi are cow i’s NE requirement (Mcal/kg DM), CP requirement (%) and daily dry matter intake (kg/cow), respectively calculated according to NRC. Path()is the value ranking as a*0.01*(vector length) and “a” is between 0 and 100. P, C1 and C2 are milk (0.35 $/kg), NE (0.07 $/Mcal) and CP (0.4 $/kg) prices, respectively. BW was calculated with Korver function as described by van Arendonk. We compared MIP with cluster (by NE) strategy in even and uneven grouping using same IOFC function, groups sizes and percentiles of NE and CP. The daily data of 471 cows from the University of Wisconsin-Madison Dairy Science herd (BW = 604.6 ± 28.9kg, DIM = 152 ± 103days, Milk yield = 43.1 ± 10.4kg/day) were used. MIP had greater average IOFC than cluster strategy in both even grouping (10.175 vs. 9.996 $/cow per d) and uneven grouping (10.383 vs. 10.193 $/cow per d). Compared with cluster strategy, MIP had 202 (42.89%) and 150 (31.85%) cows grouped differently in even and uneven grouping. FCM, feed efficiency (FCM/DMI) and IOFC were compared (Table 1). MIP approach seems to have better performance than cluster nutritional grouping, but animal experiments are needed.
Table 1. Summary of MIP and cluster grouping strategy
Scenarios | Group | MIP (low) | Cluster (low) | t | P-value | MIP (high) | Cluster (high) | t | P-value |
Even grouping (79th NE, 75th CP) | FCM | 44.5±9.1 | 40.1±7.8 | −5.6 | <0.001 | 43.5±10.7 | 47.9±10.3 | 4.5 | <0.001 |
FCM/DMI | 1.7±0.3 | 1.5±0.2 | −8.9 | <0.001 | 1.7±0.3 | 1.9±0.3 | 7.3 | <0.001 | |
IOFC | 10.3±2.8 | 9.0±2.3 | −5.3 | <0.001 | 10.1±3.2 | 11.0±2.7 | 3.3 | 0.001 | |
Uneven grouping (79th NE, 57th CP) | FCM | 44.0±9.4 | 43.8±9.5 | −0.2 | 0.9 | 44.1±12.1 | 44.5±11.7 | 0.3 | 0.8 |
FCM/DMI | 1.7±0.31 | 1.6±0.2 | −4.4 | <0.001 | 1.8±0.2 | 2.2±0.2 | 10.4 | <0.001 | |
IOFC | 10.3±3.0 | 10.2±2.9 | −0.4 | 0.7 | 10.7±3.4 | 10.0±3.2 | −1.3 | 0.2 |
Key Words: nutritional grouping strategy